You are on page 1of 38

Capacity Enhancement Core Techniques for GSM 900 and 1800

Capacity Enhancement Core Techniques for GSM 900 and 1800


Technical Memorandum

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 1 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Document History
Version 0.0 0.1 0.2 0.3 0.4 1.0 Date 27th Feb 98 3rd Mar 98 5th Mar 98 11th Mar 98 6th Apr. 98 9th Apr. 98 Author(s) S. Martin-Leon S. Martin-Leon S. Martin-Leon S. Martin-Leon S. Martin-Leon S. Martin-Leon Change Description First Draft Revision Various adition Revision Revision Revision

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 2 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

CONTENTS 1. INTRODUCTION...........................................................................................5 2. DISCONTINUOUS TRANSMISSION............................................................6 2.1 Brief description.....................................................................................................6 2.2 GSM application.....................................................................................................6 2.2.1 Implementation...................................................................................................6 2.2.2 Common channels..............................................................................................7 2.2.3 Measurements....................................................................................................7 2.3 Advantages. Effects on planning...........................................................................8 2.4 Lucents solution....................................................................................................8 2.4.1 Feature activation and parameters......................................................................8 2.4.1.1 Uplink DTX.................................................................................................8 2.4.1.2 Downlink DTX............................................................................................8 2.4.2 Measurements....................................................................................................9 3. DYNAMIC POWER CONTROL...................................................................10 3.1 Brief description...................................................................................................10 3.2 GSM application...................................................................................................10 3.2.1 Implementation.................................................................................................10 3.2.2 Common channels............................................................................................11 3.2.3 Measurements..................................................................................................11 3.2.3.1 Power control and frequency hopping.......................................................11 3.2.3.2 Quality measurements................................................................................12 3.3 Advantages. Effects on planning.........................................................................13 3.4 Lucents solution..................................................................................................14 3.4.1 Functional split.................................................................................................14 3.4.2 Process.............................................................................................................15 3.4.3 Feature activation and parameters....................................................................16 3.4.4 Algorithm optimisation.....................................................................................17 4. SLOW FREQUENCY HOPPING.................................................................18 4.1 Brief description...................................................................................................18 4.1.1 Cyclic vs. random hopping...............................................................................18 4.1.2 Baseband vs. synthesiser hopping ...................................................................18

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 3 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

4.2 GSM application...................................................................................................19 4.2.1 Implementation.................................................................................................19 4.2.2 Sequence generation.........................................................................................19 4.2.3 Common channels............................................................................................19 4.2.4 Measurements..................................................................................................20 4.2.4.1 Power control and frequency hopping.......................................................20 4.2.4.2 Quality measurements................................................................................20 4.2.5 Frequency redefinition procedure.....................................................................20 4.2.6 Mobile stations.................................................................................................20 4.3 Advantages. Effects on planning.........................................................................20 4.3.1 Frequency diversity..........................................................................................20 4.3.1.1 Description.................................................................................................20 4.3.1.2 Number of frequencies...............................................................................23 4.3.1.3 Frequency spacing......................................................................................24 4.3.1.4 Antenna diversity.......................................................................................24 4.3.1.5 Effects on planning....................................................................................25 4.3.2 Interference diversity........................................................................................25 4.3.2.1 Description.................................................................................................25 4.3.2.2 Random vs. cyclic hopping .......................................................................26 4.3.2.3 4/12 and 3/9 reuse patterns........................................................................26 4.3.2.4 1/3 reuse pattern and fractional loading.....................................................27 4.3.2.5 Multiple reuse patterns...............................................................................28 4.3.2.6 Multiple reuse patterns and fractional loading...........................................29 4.3.2.7 Fractional reuse patterns............................................................................30 4.3.2.8 Concentric cells..........................................................................................30 4.3.2.9 Control and traffic channels.......................................................................30 4.3.2.10 Effects on planning..................................................................................31 4.4 Lucents solutions.................................................................................................33 4.4.1 Base station equipment.....................................................................................33 4.4.2 Antenna coupling equipment............................................................................33 4.4.3 Fill-sender and phantom RTs..........................................................................34 4.4.4 Possible hopping configurations.......................................................................34 4.4.4.1 Limitations.................................................................................................34 4.4.4.2 Recommended configurations....................................................................35 4.4.4.3 Mixed operation of hopping and non-hopping modes...............................36 4.4.5 Feature activation and parameters....................................................................36 4.4.6 Fault defence mechanisms................................................................................37 5. CONCLUSION.............................................................................................38

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 4 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

1.

INTRODUCTION

In general there are two strategies to increase the capacity of a GSM network using its existing spectrum allocation: 1. 2. Reduce the cell size. Increase the capacity per cell.

These two strategies do not exclude each other. The second type of strategies might be used to free up some frequencies that might then be used for a microcell layer in a hierarchical network structure. Cell size reduction on its own can be very effective. It produces the greatest capacity increase: up to more than 3 times the capacity of a network coverage limited. It also improves the coverage of the network, in terms of, for example in building penetration, and cold spot elimination. However, it implies the introduction of new cell sites which has major drawbacks in terms of site acquisition and management costs. The second type of strategy provides a means to increase the capacity of the existing network in a more cost effective way. The aim of this document is to provide some insight into three of the techniques used. These, so called core techniques, are options supported by the GSM specifications and therefore, mandatory for GSM mobiles: discontinuous transmission (DTX), dynamic power control (PC), frequency hopping (FH or SFH). The first two, discontinuous transmission and dynamic power control, when used on their own, do not provide a significant capacity increase. They were devised with a different aim: to extend the mobile battery life by minimising the battery current requirements. However, when used in conjunction with frequency hopping, they provide a powerful means to increase the capacity of the network. This document analyses the advantages in terms of quality and capacity increase derived from the core techniques, their implementation using Lucents equipment is discussed, concentrating on LM4.0, and, where appropriate, guidelines for their use are given. Separate sections are included to describe each of the techniques. The benefits of their joint use are studied in the section devoted to frequency hopping.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 5 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

2. 2.1

DISCONTINUOUS TRANSMISSION Brief description

It is of common knowledge that telephone traffic is subject to alternating silence and activity periods. Typical activity factors of telephone conversations (i.e. the fraction of time a given user is actually talking) are around 0.4. It is reasonable to expect even much lower activity factors for certain data transmissions. Discontinuous transmission is a technique that takes advantage of this fact by inhibiting the transmission of the radio signal when there is no information to send (voice or data).

2.2

GSM application

2.2.1 Implementation
Discontinuous transmission, also referred to as DTX, is only relevant to some of GSM transmission modes, in particular speech and non-transparent data, simply because in the other cases (transparent data) it is difficult to assess when user data transmission can be suspended without degrading the service. With discontinuous transmission enabled, the goal is to encode speech at 13 kbit/s when the user is talking, and at a bit rate around 500 bit/s when not talking. 500 bit/s is sufficient to encode background noise so that the listener does not think that the connection is broken (this is the notion of comfort noise). The low rate of encoding results in a decrease in effective radio transmissions and therefore cochannel interference reduction. In order to implement such a mechanism, the source must be able to indicate when transmission is required. In the case of speech, the vocoder must detect whether or not there is some vocal activity. This function is called Voice Activity Detection , or VAD. At the reception side, the listeners ear must not be disturbed by the sudden silence and the decoder must therefore be able to generate some comfort noise when no speech signal is received. The discontinuous transmission mode affects the transmit operation of the mobile station and the Transcoder and Rate Adaptor Unit (TRAU or STF for Lucents equipment). The BTS is obviously concerned, but derives its behaviour dynamically from data coming from the mobile station (uplink) and from the TRAU (downlink). The distinction between comfort noise frames and speech frames can be done on the basis of the frame contents. Then the BTS decides whether to transmit the frame on the radio interface so that the minimum bit rate is met. DTX is an optional feature and must therefore be managed. Moreover, discontinuous transmission may be applied independently to each direction, so that it must take into account two components: the uplink mode and the downlink mode. The choice of the strategy for applying discontinuous transmission is one of the many configuration parameters which operators may use to optimise their network. Several considerations must be taken into account in this strategy. For instance, GSM mobile to mobile calls suffer a loss in quality when discontinuous transmission is applied to both radio segments (double clipping). The operator may therefore choose not to apply the downlink discontinuous transmission mode, if MS-to-MS call numbers are significant.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 6 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

As far as the downlink discontinuous transmission mode is concerned, it must be established on a connection by connection basis by the Mobile Switching Centre (MSC). The MSC tells the Base Station Controller (BSC), which configures the Base Transceiver Station (BTS), itself configuring the TRAU. Means to change the downlink discontinuous transmission mode on an established connection also exist, when the transmission mode changes. The mobile station can be ordered to use the discontinuous transmission mode in the uplink direction as a cell option. The cell options are regularly broadcast on the BCCH for mobile stations in idle mode, and they are also part of the general information sent to mobile stations on their SACCH when they are in dedicated mode. The information sent on the SACCH is specified by the BSC on a transmitter/receiver (TRX) basis. It can have three values: DTX must be applied, must not be applied or may be applied. The decision as to whether the uplink DTX feature is to be activated for a particular call, is determined by the mobile station on advice from that information.

2.2.2 Common channels


One peculiarity related to common channels is that the frequency they use for the downlink must be emitted continuously and at full power, even if no information needs to be conveyed on some bursts. This is needed because mobile stations in neighbouring cells continuously perform measurements on this frequency. In idle mode it is used to determine the best cell they should listen to. In dedicated mode it is done to report measurements for hand-over preparation, even when the different base stations are not synchronised. When there is no information transfer request, a specific pattern is emitted (the fill frames).

2.2.3 Measurements
When discontinuous transmission is applied, some slots belonging to a channel may not be used for transmission. This is indeed the goal of it, but then measurements on these slots will obviously report a low reception level, and a bad quality. To avoid this problem, the GSM Specifications impose that at least 12 bursts are sent within each reporting period (SACCH superframe). These bursts amount to the systematic use of the SACCH (4 bursts constituting a coding block) and 8 bursts on the TCH itself. For speech, these bursts contain silence description frames (SID frame). In addition to this minimum transmission rule, the Specifications require the BTS and the mobile station to report two sets of measurements concerning the connection: full measurements, done on all slots which may be used for transmission in the reporting period, sub measurements, done only on the mandatory sent bursts and blocks. Finally, both the BTS and mobile station report for each measurement period whether discontinuous transmission was used or not, in other terms, whether all bursts were transmitted, thus enabling the processes using the measurements (power control and handover) to discard the full measurements when discontinuous transmission has been used. Due to the reduced number of input values for the averaging process, the results based on the sub measurements are less accurate (reception level is averaged on 12 bursts instead of over 100 bursts). That specially affects the quality measurements, which are based on estimated error probabilities before channel decoding, and therefore more sensitive and statistically unreliable in the case of subset measuring, than the received level.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 7 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Such unreliability can result in an increase in the number of dropped calls, as it has been reported in field measurements. Therefore, for those processes in need of quality, some kind of solution for this problem has to be provided when discontinuous transmission is used.

2.3

Advantages. Effects on planning.


The use of DTX has two main advantages. It decreases power consumption in mobiles thereby enhancing handset battery life. It reduces the overall interference in the system. This implies a quality increase. However, this quality increase cannot be translated into a capacity increase since system planning has to be done for worst case situations. Interference with DTX has an on/off nature with the maximum interference being the same as without discontinuous transmission, and the rate of switching (rate of activity and inactivity periods) is not high enough for coding and interleaving to be effective in averaging the variation.

In terms of quality increase, DTX has an advantage when compared with dynamic power control: the interference reduction is random whereas in the case of power control it depends on the geographical location of the mobiles.

2.4

Lucents solution

Lucents equipment offer the possibility to use discontinuous transmission both in the uplink and downlink for speech and non-transparent data communication.

2.4.1 Feature activation and parameters 2.4.1.1 Uplink DTX


Uplink discontinuous transmission in a certain cell (BTS) can be deployed by setting, in the Base Transceiver Station window of the OMC GUI, the parameter Uplink in the DTX subsection (AUI parameter DTX of the managed object class BTS) accordingly. It can have three values: May be used (0). Shall be used (1). Shall not be used (2). This parameter is then used to set the value of the DTX field in the cell options information element broadcast in the BCCH and in the associated SACCH of a communication.

2.4.1.2 Downlink DTX


Downlink DTX can be set separately for speech (system release 6.5 feature) and for nontransparent data (system release 6.7.2 feature). To do this the corresponding parameters have to be set in the BSS and the InterWorking Function (IWF) subsystems.

Speech
This feature may be either enabled or disabled on a per BTS basis via the OMC by setting, in the Base Transceiver Station window of the OMC GUI, the parameter Downlink Speech in the DTX

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 8 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

subsection (in the AUI DownlinkDtx attribute of the managed object class BTS contains the boolean downlinkDtxSpeech ). The default setting is disabled (false). In the MSC the switch option Downlink DTX Mode in the WBOPM (Wireless Base Office Parameters Miscellaneous) view, can be enabled and disabled in the corresponding windows of the Recent Change and Verify program (RC/V). This applies to all the BSS supported by one MSC. Its default mode is disabled. DTX is then permitted for the connection if DTX is requested by the MSC and enabled by the OMC.

Data
This feature may be either enabled or disabled on a per BTS basis via the OMC by setting, in the Base Transceiver Station window of the OMC GUI, the parameter Downlink Data in the DTX subsection (in the AUI DownlinkDtx attribute of the managed object class BTS contains the boolean downlinkDtxData). The default setting is disabled (false). If this attribute is set, the BTS acts according to the DTX commands issued by the IWF in the received RLP frames. To set DTX in the IWF, the IWF includes an option: DTX Mode, which can be set by changing the value in the IWF-2 menu. Its default mode is disabled.

2.4.2 Measurements
In order to overcome the inaccuracy of measurements with discontinuous transmission the following process is envisaged. In the BSC a linear unweighed sliding window averaging is performed for all radio link measurements. In the case of RXQUAL measurements, the sliding window depth is A_QUAL_RR (where RR can be PC for power control and HO for handover). A measurement where no discontinuous transmission has been used is shifted into the averaging window W_QUAL_RR times, whereas the number of times is just one for measurements with discontinuous transmission. If the discontinuous transmission flag has not been received, The sub measurements are also chosen and shifted into the averaging window once. In this way, more accurate measurements are given a higher weight in the resulting averaged value, which is then used in the power control and handover processes. This weighing process only applies to RXQUAL measurements. RXLEV measurements are shifted into the averaging window once no matter whether they are full or sub measurements. basis. Both A_QUAL_RR and W_QUAL_RR are parameters that can be set via the OMC on a BSC

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 9 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

3. 3.1

DYNAMIC POWER CONTROL Brief description

Dynamic power control refers to the possibility to modify the transmission power over the air, dynamically during a connection, independently both for the mobile station and the base station. Control of the mobile station power has been used in most of the existing cellular systems, in order to save battery power in the terminals (power consumption). However, if the received signal level is too high, by reducing the transmission power while keeping a satisfactory quality level on the communication, the interference caused on other calls in surrounding areas is decreased. Since co-channel interference is one of the main capacity limiting factors in a cellular system, power control can also be used to improve spectral efficiency. Basically, in the literature, two different types of algorithms for power control have been proposed for any kind of cellular system. One is based on the principle that the power should be reduced as the path loss is decreased. The simplest, and most used, of this type of algorithm keeps the received signal strength constant. It has been shown that such an approach gives very little gain in capacity. A variation of this scheme is based on received signal strength, but the change in path loss is only partly compensated for. For this latter scheme it has been shown that there is some gain in capacity. In the second type of algorithm the focus is instead on quality. Since most of the connections experience an excess in quality (i.e. C/I ratio), it seems natural that the power should be controlled according to the quality of the call. In this way, the unnecessarily high C/I margin found in most calls can be converted into capacity. Analysis has shown that a large capacity gain is obtained by using a power control scheme giving the same quality to all users (roughly 5-7 dB interference level reduction in threshold value). The problem with such an algorithm is that it needs centralised control. Thus it is, in its present version, of little practical use in a real cellular environment. Different distributed algorithms have been devised, most of them showing better capacity results than the ones based on signal level. Mixed algorithms, where both the signal level and the quality are taken into account, are also possible.

3.2

GSM application

3.2.1 Implementation
In GSM, both uplink and downlink power control may be applied independently from each other; furthermore they are applied independently for each mobile station. The range specified for uplink power control lies between 20 and 30 dB, in 2 dB steps, depending on the mobile station power class. The range used for downlink power control is manufacturer dependent and may be up to 30 dB, also in 2 dB steps. The control of the transmission power is a network option, i.e. the operator may choose to apply it or not, in one direction, or in both. All mobile stations though, must support the feature. Power control on both directions is managed by the BSS. The transmission power of the mobile station is chosen by the BSS, and the commands to regulate it are issued to the mobile station.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 10 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

The BSS computes the required MS transmission power through reception level measurements performed by the BTS, taking into account the MS maximum transmission power as well as quality measurements done by the BTS. This last parameter helps to ensure that transmission quality is kept above some acceptance threshold. For the downlink direction, the BTS transmission power is also computed by the BSS for each connection, based on the measurements performed by the mobile station and reported regularly to the BTS. Inside the BSS, the split of tasks between BTS and BSC is basically an option for the manufacturer. The specification of the A bis interface is basically adapted to the implementation of power control on the BSC, but implementation in the BTS is also possible. At the start of a connection, the initial value of the transmission power (both for mobile station and BTS) is chosen by the BSC. In the case of an initial assignment, the information available to choose this power is at best very small. Therefore, in GSM, the initial power level to be used by a mobile station for the first messages sent on the new dedicated channel is fixed on a cell-per-cell basis, and is the same level as used for sending random access bursts. The value of this level is broadcast on the BCCH, to be known by all mobile stations before any access attempt. A mobile station whose power level is below the broadcast value shall simply use its maximum power level instead. Except at the start of a channel connection, a command to change the transmission power does not trigger an immediate transition to the ordered value in the mobile station. The maximum variation speed is of 2 dB each 60 ms. That means that a high jump in the power control commands will be answered gradually.

3.2.2 Common channels


See section 2.2.2.

3.2.3 Measurements 3.2.3.1 Power control and frequency hopping


The combination of power control with frequency hopping using the BCCH carrier raises a problem about measurement accuracy in traffic channels using such combination. Power control can be applied on any frequency except for the BCCH frequency, which must be transmitted with a constant power in the downlink. The result for the channels under consideration is that power control applies only to a subset of the bursts, whereas other bursts (those using the BCCH frequency) are sent with a fixed transmission power. This could lead to inaccurate reception level measurements. In order to alleviate this problem, the mobile station is requested in such cases not to take into account the slots falling on the BCCH frequency in the reception level estimation. This is controlled by an indicator, the PWRC indicator, sent on a connection basis to the mobile station when the following conditions are met: the channel hops on at least two different frequencies, one of those frequencies is the BCCH frequency and downlink transmission power control is in use. This problem does not apply to quality measurements which are performed as usual.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 11 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

3.2.3.2 Quality measurements


In the GSM system the only way to get information about the current interference level, to be used in the power control algorithm, is the received quality (i.e. RXQUAL) which is related to the C/I ratio. However, according to the NTC, the eight available quality levels (RXQUAL=0..7) represent a C/I interval which in the worse case can be as small as 9 dB 1 (see Figure 1). Outside this interval the level of interference cant be determined.
RXQUAL 7

0 9 - 18 dB

CIR

Figure 1 Qualitative relationship between RXQUAL and CIR Because of the small C/I window which can be measured by RXQUAL, the direct C/I control would cause harmful quality problems in combination with shadow fading. For example when the power control has reached a target of RXQUAL=2, an attenuation increase caused by shadowing would break down the quality immediately (e.g. to RXQUAL=6). The power control wont be able to react fast enough, because it has not been able to detect the lowering of the quality until it is too near the level where it deteriorates to unacceptable values. Therefore pure C/I control is impossible in the current GSM system. Another point to be taken into account is the different mappings of C/I to RXQUAL on TCH channels depending on the propagation scenario. Because RXQUAL represents the estimated error probabilities before channel decoding, it does not consider the varying efficiency of coding, interleaving and bit error correction under different environmental conditions. The following figure shows the result of simulations using the TRASIM simulation tool to model the GSM receiver.
25.6% 12.8% 6.4% 3.2% 4 1.6% BER -> 0.8% 0.4% 1 0.2% 0 0.1% 0 5 10 15 CIR [dB] -> 20 3 RXQUAL -> 2 TU3 no FH TU3 with FH TU50 no FH TU50 with FH RA250 no FH 7 6 5

Figure 2 C/I to RXQUAL mapping for different scenarios


1

This case happened at receiver sensitivity level for a static channel model. RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 12 of 40

Author : Date:

Capacity Enhancement Core Techniques for GSM 900 and 1800

TU3 and TU50 refer to a typical urvan environment and mobile stations with a speed of 3 and 50 km/h respectively. RA250 refers to a rural area environment and a mobile at 250 km/h.

3.3

Advantages. Effects on planning

The following figure shows the C/I ratio perceived by a mobile station as a function of the distance to the base station normalised to the distance between interfering base stations. It can be seen that there is a lot of power wasted when the mobile is near the base station to ensure that, when it is near the cell border, the C/I ratio and hence the quality of the communications, remains good enough.
70 60 50 40 C/I 30 20 10 0 0 -10 d/D 0.2 0.4 0.6

Wasted power

Cell edge Target C/I

Figure 3 C/I ratio as a function of the normalised distance In a system where all mobiles transmit at maximum output power, the total interference is greater than if some mobiles regulate their power. At the cell border, the mobiles transmit at maximum power in both regulating and non regulating systems. Hence in the non regulating system, the base station receives lower C/I values for the mobile near the cell border than in the regulating system. The mobiles near the cell border also produce the lowest C/I value in both systems. When regulating, the mobiles near the base station will transit at lower power and hence signal strength will be lower than in the non regulating system. In some cases that will result in lower C/I ratios (when the interferer is at the cell border). However, these mobiles already have good conditions, i.e. they are in the upper region of the C/I curve, and do not suffer from degraded performance. These results can be better understood with the following figure. In it the C/I ratio is plotted against the distance between the desired mobile and its base station normalised to the distance between interfering base stations, without and with downlink power control. In the latter case the C/I ratio depends on the position of the interfering mobile station. Two different cases and the average value are shown.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 13 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

70 60 50 40 C/I 30 20 10 0 -10 0 0.2 d/D 0.4 0.6 PC, average PC, interf. at cell border PC, interf. near base-station w ithout PC

Figure 4 C/I ratio as a function of the normalised distance Therefore the use of power control provides an increase in the global quality of the system (less calls in interference conditions) which can be translated into a capacity increase by planning the network with lower average C/I values2. The gain is not enough to be able to jump from a 4/12 reuse factor to a 3/9 reuse factor, but its effect might be noticeable when using automatic planning tools that take power control into consideration.

3.4

Lucents solution

Lucents equipment implements level and quality based dynamic power control. The multi-step threshold comparison technique proposed as an option in GSM Rec. 05.08, Appendix A, Section 3.2, is not supported. Instead, a similarly powerful technique is provided which uses sliding window measurement averaging in conjunction with a single-step threshold comparison process.

3.4.1 Functional split


follows: The functional split for the overall power control process in Lucents BSS subsystem is as

BTS: All measurement results necessary for Power Control, i.e. both the downlink measurement results reported by the MS, and the uplink measurement results reported by the RT, are acquired by the corresponding BTS, and transmitted to the BSC without any preprocessing. BSC: Measurement averaging (pre-processing) as well as the entire uplink and downlink power control processing is performed in the BSC. This is done by independent processes for all TCH and SDCCH channels. In particular, for each dedicated channel, uplink and downlink processes operate independently of each other. The following description of the process and parameters relates to LM4.0. Differences with LM5.0 are also highlighted.
2

Something very similar happens with interference diversity. See 4.3.2.1. RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 14 of 40

Author : Date:

Capacity Enhancement Core Techniques for GSM 900 and 1800

3.4.2 Process
The power control algorithm implemented can be split into the following basic steps: 1. PC measurement averaging: both RXLEV and RXQUAL measurements are averaged using average window sizes and weighting factors: AV_LEV_PC, A_QUAL_PC, W_QUAL_PC. 2. PC threshold comparisons: the averaged level and quality measurements are compared with upper and lower limits and a decision is made on whether an increase or decrease in transmit power is required (Figure 5).
RXLEV -48 dBm (63) 1 U_RXLEV_XL_PC 2 L_RXLEV_XL_PC -110 dBm (0) 7 L_RXQUAL_XL_PC U_RXQUAL_XL_PC 0 1 - Range of operation without Power Control 2 - Target range ofoperation with Power Control

RXQUAL

Figure 5 Operation of Power Control 1. PC execution: in the case the actual transmit power is not at its minimum or maximum value, a fixed step power increase or decrease is ordered. 2. PC disabling : the total time between a new power command and the effect in the measurements is 2-3 SACCH multiframes in the downlink and 3-4 SACCH multiframes in the uplink. Because of the delay caused by the dead time and the process of averaging measurements, the effect of the command takes some time to be noticed by the BSC, and a power change command could be sent after a previous power change command, even when the power is set to the correct value. This decreases the stability of the power control loop. To avoid incorrect power commands, power command disabling is used. The power control algorithm waits for a power command acknowledge, and after this, an extra interval of time to ensure the power control command has started influencing measurements. If the first timer expires before the acknowledgement has been received, the process will consider the current transmit power the one that should be set at the moment. It will wait for the extra interval and then the threshold comparison process will be resumed. This process takes into account the different transmitter characteristics of the systems GSM 900 and GSM 1800. The dynamic range of Lucents BTSs is 30 dB.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 15 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Handovers only take place when the mobile is transmitting at its maximum power, except in the case of hierarchical structures. In this case, the power control process varies in that first it has to be decided whether a handover of a power control step is the right alternative. In addition to the regular PC process, a procedure is implemented which immediately directs MS and BTS to use their maximum power, if: there is a Radio Link Failure warning (Such a warning is produced by the BTS as soon as it detects that a radio communication on the uplink is about to break-down); the radio link control process is started while power control is disabled; the allocated channel is on the BCCH carrier of a non-hopping system; a mandatory handover in the lower cell layer of a hierarchical cell structure can be prevented. In LM5.0, the threshold comparison and execution processes are modified to include a variable step size for emergency power control, which is used to influence the recovery time of the PC loop. In an emergency, i.e. when the receive level measurement (not the averaged one) is well outside the normal range of operation, a step size is calculated so that receive values are directed in one step towards that range.

3.4.3 Feature activation and parameters


Power control for communications through a certain BTS can be enabled in the downlink and uplink independently by setting parameters EN_MS_PC (uplink) and EN_BS_PC (downlink) of the POWER object associated with the BTS accordingly. Before doing that the following parameters should have been set to their proper values: Maximum transmit power values: MS_TXPRWR_MAX: maximum TX power a MS is permitted to use on a dedicated control channel or a traffic channel within the serving cell. Averaging measurement parameters: A_LEV_PC: averaging window size for receive power level measurements. A_QUAL_PC: averaging window size for quality measurements. W_QUAL_PC: weighting of full-set quality measurements with respect to sub-set quality. Threshold levels: L_RXLEV_UL_P, U_RXLEV_UL_P, L_RXQUAL_UL_P, U_RXQUAL_UL_P: uplink lower (L) and upper (U) RX_LEV and RX_QUAL threshold. L_RXLEV_DL_P, U_RXLEV_DL_P, L_RXQUAL_DL_P, U_RXQUAL_DL_P: the same for the downlink. Power step sizes: POW_INCR_STEP_SIZE, POW_RED_STEP_SIZE: step sizes used when increasing or decreasing the MS and BTS transmit power. Timer values: P_CON_ACK: power control acknowledge time. P_CON_INTERVAL: minimum interval between successive modification of the radio frequency power level. The following table shows recommended values.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 16 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

O&M parameter A_LEV_PC A_QUAL_PC W_QUAL_PC L_RXLEV_DL_P L_RXLEV_UL_P L_RXQUAL_XL_P U_RXLEV_DL_P U_RXLEV_UL_P U_RXQUAL_XL_P POW_INCR_STEP_SIZE POW_RED_STEP_SIZE P_CON_ACK P_CON_INTERVAL

Range 1-31 (SACCH multiframes) 1-31 1-3 0-63 (0-110 dB;63-48 dB) 0-63 0-7 0-63 0-63 0-7 0-2 (2-6 dB) 0-1 (2-4 dB) 0-31 (2 SACCH multiframes) 0-31

Default value 6 6 1 (if no DTX is used) 25 25 3 35 35 1 2 (6 dB) 1 (4 dB) 4 4

3.4.4 Algorithm optimisation


Presently, work is being undertaken within Lucent in order to specify an optimised power control algorithm that improves the performance of the current one. Its stability for certain range of the parameters and its speed to adapt to a fast changing environment will be enhanced. Also new methods are being studied to better estimate the C/I value with the range of RXQUAL available in GSM.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 17 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

4. 4.1

SLOW FREQUENCY HOPPING Brief description

Frequency hopping consists of changing the frequency used for transmission at regular intervals. In GSM a frequency hop takes place every new burst (every 4.615 ms), resulting in 217 hops per second. This is known as slow frequency hopping (SFH) as opposed to fast frequency hopping, where there are several hops per symbol transmitted. SFH has been introduced in GSM mainly in order to deal with two specific problems which affect the transmission quality: Multipath fading: The radio signals are subject to multipath fading, which is space and frequency selective. A slowly moving mobile may stay in a fading dip long enough to suffer severe information loss. Frequency hopping combats multipath fading by exploiting its frequency selectivity: changing frequencies also means changing fading patterns. Introducing frequency diversity will, together with the interleaving and coding, improve the transmission quality of the link, for slow moving users in particular. Interference : Without frequency hopping, strong signals from neighbour cells transmitting on or close to a carrier frequency will affect the carrier signal continuously, which may have a negative effect on the transmission performance. Frequency hopping can cause different signals to interfere with the carrier at different times, a property called interference diversity . At a system level, the result is a smearing of the interference levels between users, an effect often called interference averaging. Furthermore, interference diversity gives a second effect: since consecutive bursts of information are received under different interference conditions, the risk of a sequential information loss is reduced. This, together with the interleaving and coding, will improve the transmission link quality.

4.1.1 Cyclic vs. random hopping


The hopping can be either cyclic or random. Cyclic hopping means that all mobiles use frequencies consecutively from the set allocated to the cell, e.g f1, f2, f3, f4, f1, f2, . With random hopping, mobiles use uncorrelated pseudo-random hopping sequences. In the case of a regular assignment pattern, the probability of two mobiles in interfering cells using the same frequency in the same time slot is 1/N, where N is the number of frequencies assigned to each cell, e.g. MS1: f1, f4, f4, f2, f1, f3, ; MS2: f2, f1, f4, f3, f2, f1, .

4.1.2 Baseband vs. synthesiser hopping


There are two methods to produce frequency hopping by the base station: baseband hopping and synthesiser hopping. Baseband hopping is realised by switching the output from each of the baseband processing sections between the inputs of the RF portions of each RT. In this way, the RTs do not need to retune, but each channel effectively hops over the available frequencies. Author : Date: RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 18 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

The primary limitation of this type of frequency hopping is that the maximum number of hopping frequencies equals the number of RTs in that particular cell. Synthesiser hopping implements frequency hopping by retuning the RF portion (transmit and receive) of each TRX (RT) in a defined hopping pattern. Therefore, the output of each baseband processing section is always connected to the same RF portion (unlike in baseband hopping). This allows each RT to hop over as many frequencies as desired, independently of the number of RTs in the cell. However wideband hybrid combiners must be used with this type of hopping, and not filter combiners, which take some time to change their frequencies. Hybrid combiners have much higher insertion losses than filter combiners and, therefore, the number of radios is limited in a cell.

4.2

GSM application

4.2.1 Implementation
In GSM with frequency hopping every mobile transmits its time slots according to a sequence of frequencies that it derives from an algorithm. The frequency hopping occurs between time slots and, therefore, a mobile station transmits (or receives) on a fixed frequency during one time slot and then must change frequency before the time slot on the next TDMA frame.

4.2.2 Sequence generation


For a set of N frequencies, up to 64xN different hopping sequences can be built. They are described by two parameters: the MAIO (Mobile Allocation Index Offset) which may take as many values 1 ..N, the HSN (Hopping Sequence Number) which may take 64 different values. Two channels bearing the same HSN but different MAIO never use the same frequency on the same burst. Two channels using the same frequency list and the same time slot, but bearing different HSN, interfere 1/Nth of the bursts, as if the sequences were chosen randomly. The sequences are indeed pseudo-random, except for the special case of HSN=0, where the frequencies are used one after the other in order (cyclic hopping). Channels in one cell using the same set of frequencies bear the same HSN and different MAIO to avoid interference between channels inside a cell.. If random hopping is used, different HSN should be used in distant cells using the same frequency set, in order to gain from interference diversity.

4.2.3 Common channels


Common channels (FCCH, SCH, BCCH, PAGCH and RACH) must use a fixed frequency. This constraint is meant to ease initial synchronisation acquisition. Similarly, extension sets of common channels are also forbidden from hopping and use the same frequency as the primary group, so that there is no need to transmit the description of their frequency organisation on the BCCH or time slot 0.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 19 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

4.2.4 Measurements 4.2.4.1 Power control and frequency hopping


See section 3.2.3.

4.2.4.2 Quality measurements


As it was already said in section , the mapping of RXQUAL to C/I varies depending on the propagation environment. In the case of frequency hopping, when its interference diversity effect is used (see 4.3.2.2), the mapping also varies call to call, depending on the changing interference conditions. That means that RXQUAL is not a reliable measurement of the quality of the connection when used in such conditions.

4.2.5 Frequency redefinition procedure


In dedicated mode and group transmit mode, this procedure is used by the network to change the frequencies and hopping sequences of the allocated channels, when the network frequency plan is changed, so that as few calls as possible are disturbed. The network sends to the MS a FREQUENCY_REDEFINITION message containing the new parameters together with a starting time indication. When receiving such a message, the MS modifies the frequencies/hopping sequences it uses at the exact indicated time-slot, i.e. the indicated time slot is the first with new parameters. All other functions are not disturbed by this change. New parameters can be the cell channel description, the mobile allocation and the MAIO. Other parameters describing the allocated channel are identical to the current parameters. Despite this procedure, some calls may be lost. This can happen when the MSC asks for a channel for handover, and the request is acknowledged with the actual channel information. If afterwards, a redefinition procedure is started for this channel, and the MS is handed over to that channel at the same time, the call is lost.

4.2.6 Mobile stations


There are certain mobiles which cannot cope with the redefinition procedure. It has also been reported that others are not able to hop on the SDCCH channels or have problems when using SFH in conjunction with DTX downlink and/or dynamic power.

4.3

Advantages. Effects on planning

4.3.1 Frequency diversity 4.3.1.1 Description


Multipath fading is speed and frequency dependent. For speech services, a speed of 35 km/h in the 900 MHz band (17.5 in the 1800 band) is enough to overcome its effects. For more slowly moving users, the GSM error correcting mechanisms are not enough . However, using frequency hopping the same performance can be obtained.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 20 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Figure 6 shows the necessary C/N o ratio as a function of vehicle speed at fixed frequency allocation (900 MHz band) and with the use of an ideal frequency hopping for a bit error ratio (BER) of 0.5% which is regarded as tolerable for speech transmission.
13 12 11 10 9 8 7 6 5 4 0 50 100 v [km /h] 150 200 C/No [dB] for BER=0.5%

Without FH FH

Figure 6 Required C/No against vehicle speed for a BER of 0.5% With ideal frequency hopping, the best possible transmission quality is obtained at almost every vehicle speed. A slight degradation is observed at very high vehicle speeds. The reason for this is a significant change in the multipath profile even within one time slot that cannot be solved by the equaliser. A similar situation is observed for co-channel and/or adjacent-channel interference. Figure 7 contains an impression of the necessary C/I ratio in terms of the current vehicle speed. The dependence is even more clearly marked than for noise interference, as here the power of the interference signal also fluctuates in accordance with speed.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 21 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 22 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

16 C/I [dB] for BER=0.5% 14 12 10 8 6 0 50 100 v [km/h] 150 200 Without FH FH

Figure 7 Required C/I against vehicle speed for a BER of 0.5% The two causes of non-ideal frequency hopping are: a low number of frequencies and a frequency spacing that is too small. They will be studied in the following sections.

4.3.1.2 Number of frequencies


The decisive factor for the minimum length of the hopping period for ideal frequency diversity is that a different frequency has to be used for every slot inside the interleaving block. In the case of cyclic frequency hopping the hopping period must be at least as long as the interleaving depth (8 for speech), and a greater period will not cause additional gain. In the case of a shorter hopping period at least two of the time slots, over which a code word is spread, are transmitted at the same frequency so the fading processes are strongly correlated for them at a low velocity. That will lower the gain. The bit error curves in Figure 8 provide an impression of the losses to be anticipated at low vehicle speed (5 km/h). It is notable that hopping over only 4 frequencies comes 1 dB closer to ideal hopping.
1

0.1 BER

Without FH 2 freqs 4 freqs

0.01

8 freqs

0.001 0 2 4 6 8 10 12 14 C/No [dB]

Figure 8 Effect of number of frequencies on bit error ratio (BER), v=5 km/h For random hopping, the probability of using the same radio frequency channel within the interleaving depth is depth/N, where N is the number of frequencies in the sequence. That means that the fading decorrelation within one interleaving block is never optimal. The following table shows the Author : Date: RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 23 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

C/No required for a BER of 10-2 in the case of using cyclic and random frequency hopping for different number of frequencies. It can be seen that the frequency diversity gain with 8 frequencies is 1-2 dB lower for random hopping. Cyclic SFH C/No for Gross C/No for Class 1 FER=2% BER=10-2 Level Gain Level Gain [dB] [dB] [dB] [dB] 9.5 0.0 11.5 0.0 7.0 2.5 8.5 2.0 6.0 3.5 7.5 3.0 5.0 4.5 6.5 4.0 4.0 5.5 5.5 5.0 4.0 5.5 5.5 5.0 Random SFH C/No for Gross C/No for Class 1 FER=2% BER=10-2 Level Gain Level Gain [dB] [dB] [dB] [dB] 9.5 0.0 11.5 0.0 7.5 2.0 9.5 2.0 6.5 3.0 8.5 3.0 6.0 3.5 8.0 3.5 5.5 4.0 7.5 4.0 5.0 4.5 7.0 4.5

No. of frequencies 1 2 3 4 8 12

The results in this table cannot be directly compared with the results of the previous figures because different propagation conditions are used (in particular typical urban). Measurements indicate that the gains realised from frequency hopping can be smaller than predicted due to the diminished severity of multipath propagation when compared to flat fading. In normal environments the different paths arrive a different times so the cancelling is not complete. With respect to interference, the same statements apply.

4.3.1.3 Frequency spacing


The frequency spacing has to be large enough to ensure that different frequencies suffer uncorrelated fading. For example, in a typical urban environment (TU) a channel separation of 400600 kHz (2-3 GSM channels) would be needed. In most environments coherence bandwidths of less than 1 MHz can be expected, and therefore 1 MHz (5 GSM channels) can be recommended as the minimum frequency spacing for outdoor scenarios Indoor systems, however, are generally characterised by large coherence bandwidths which indicates that a lower frequency diversity gain would be achieved than in outdoor systems with the same hopping bandwidth. On the other hand, indoor users usually have lower velocity than outdoor users indicates potential for a higher frequency diversity gain with frequency hopping. Simulations have shown that even though the gain achieved is smaller (for a 2% FER 1.7 to 3.3 dB gain in C/N o instead of the 5 dB in a TU scenario with 5 MHz hopping bandwidth), it is still significant.

4.3.1.4 Antenna diversity


Another method generally used to overcome the effects of multipath fading is antenna diversity. This technique also achieves gains in conjunction with channel encoding and interleaving, but, since it uses space diversity, the gain is independent of vehicle speed. Both methods can be used together to provide a further considerable increase in gain. The total gain, however, does not amount to the sum of the individual gains. The following table will illustrate this effect:

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 24 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

No diversity Ideal diversity

C/No for FER=2% No SFH Ideal SFH Level Gain Level Gain [dB] [dB] [dB] [dB] 12.5 0.0 5.5 7.5 5.8 6.6 1.8 10.7

C/I for FER=2% No SFH Ideal SFH Level Gain Level Gain [dB] [dB] [dB] [dB] 15.5 0.0 7.3 8.2 8.0 7.5 3.2 12.2

It must be noted that the antenna diversity gain is very high for TU environments, but it drops in other test conditions (rural area or hilly terrain).

4.3.1.5 Effects on planning


Planning of a radio network must be based on the unfavourable transmission conditions at slow vehicle speed. Frequency hopping offers a possibility to compensate for the degradation in transmission quality, making it largely independent of vehicle speed. That could mean that the smaller C/No and C/I values for medium speed vehicles can be used in the planning process of areas where a meaningful percentage of pedestrian users are expected. The reduction of the required C/N o that frequency diversity provides cannot be translated into a reduction in sensitivity values because the BCCH carrier cannot hop. In terms of interference reduction, something very similar happens. The improvement in the required C/I levels cannot be translated into a tighter reuse pattern in the BCCH carriers. In this case, however, TCH carriers can be planned differently, i.e. different C/I requirements can be set in the frequency planning process (use TU50 values instead of TU3, for example), and frequency diversity can be used as a means for increase capacity. The increase will largely depend on the number of frequencies that are set in the hopping sequences, which as seen have an effect on the required C/I value. Care must be taken to ensure that the separation between frequencies assigned to a cell is appropriate to the propagation environment. If possible, the BCCH frequency shall be included in the hopping sequences for those channels that do not occupy time-slots assigned to control channels. In general, just by switching on frequency hopping in an existing network with the current frequency plan, an increase in the quality perceived by slow moving users will be noticed.

4.3.2 Interference diversity 4.3.2.1 Description


Interference diversity is a property that is generally associated Spread Spectrum systems, and GSM can also take advantage of it by using frequency hopping. In high traffic areas, such as large cities, the capacity of a cellular system is limited by its own interference caused by frequency reuse. Since the aim of a system is usually to satisfy as many customers as possible, its maximum capacity is calculated based on a given small proportion (generally around 10%) of calls subject to a noticeable decrease in quality due to interference. Because of this concept of worst case, the capacity of a system is better when the statistical spread around this mean value is as small as possible. That can be seen in the following figures. In the first one (Figure 9, left) we can see the C/I distribution for

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 25 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

systems with equal average C/I value but different deviation. A smaller deviation means that you can plan the system for a lower average C/I value without loss in quality (Figure 9, right)
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 20 25 C/I [dB] 30 35 40 45 50 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 20 25 C/I [dB] 30 35 40 45 50 12 dB 7 dB

STANDARD DEVIATION
12 dB 7 dB

Figure 9 Example of C/I distributions In a system where the interference level perceived depends on the mean of interference level averaged among all the time-slots, which is the case of GSM thanks to its channel encoding and interleaving, the spread reduction in the interference suffered by a call can be achieved by changing the interferers, and therefore the interference, slot by slot. The greater this number of interferers for a given total mean, the better the system. This is how interference diversity operates. Interference diversity also allows the system to take advantage of the fact that traffic peaks do not occur at the same time in all cells. The averaging effect will spread the interference caused by the highly loaded cells among the rest, causing only a small degradation in the quality of the system. Frequency hopping produces interference diversity if it is ensured that the interference in each of the hopped frequencies is different. Also, the greater the number of hopping frequencies, the greater the gain, because the interference is averaged among more interferers.

4.3.2.2 Random vs. cyclic hopping


Traditionally interference diversity has been connected with random frequency hopping. That is because the effect of SFH has been studied in regular frequency assignment pattern scenarios (4/12, 1/3, ). In them, with random frequency hopping, cells use uncorrelated pseudo-random hopping sequences that make the interference vary randomly for each burst transmitted. If cyclic hopping were to be used in such scenario, the same hopping frequencies, and the same hopping pattern, would be used in the next cochannel cells, and there would initially be no advantage in terms of interference diversity. However, if there is a different frequency reuse pattern for every set of hopping frequencies, the sources of co-channel interference will switch from time slot to time slot, even if cyclic hopping is used. Thus there is an interference diversity effect no matter the kind of hopping used. Accordingly, two kinds of strategies can be followed: tightening frequency reuse factor with random hopping using multiple reuse factor patterns with both random and cyclic hopping.

4.3.2.3 4/12 and 3/9 reuse patterns

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 26 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Simulation results show that frequency hopping in the traffic carriers used with a 4/12 reuse pattern increases the quality in the network due to a reduction in interference, only if used in conjunction with downlink discontinuous transmission, and downlink power control. The reason for this is that in the downlink, frequency hopping on its own produces little benefit. All potential interferers from any given cell (base stations) are at almost the same distance, due to the regular cell structure. Consequently they have similar transmission paths and there is little interference variation. The benefit on the mobile to base link is achieved due to the geographical distribution of interferers. Discontinuous transmission and power control produce the necessary variation. The increase in quality can be traded with capacity by tightening the reuse factor to 3/9. These results were obtained with 100% traffic load. In reality such load is never achieved because networks are planned for a certain percentage of blocking during busy hours. In practise different interference conditions between slots due to traffic can then be expected in the downlink. Irregular propagation conditions will also affect the interference conditions, enabling the downlink to benefit from frequency hopping even if power control and DTX were not used.

4.3.2.4 1/3 reuse pattern and fractional loading


The capacity of a GSM network is limited by either the number of traffic channel (hard blocking) or the interference from the neighbour cells (soft blocking). It can be seen that, from a hard blocking point of view, small reuse factors give a superior performance to higher reuse factors, due to the so called trunking efficiency. The small reuse factors will though be limited by soft blocking, i.e. interference, and will not be able to accept more than a given amount of traffic. It is therefore expected that the maximum capacity will be somewhere between a high reuse factor and a small reuse factor. In the case of interference diversity the soft blocking limit will be a lot lower, allowing more traffic load for a given reuse factor. To illustrate this, various reuse schemes with frequency hopping in the traffic carriers have been simulated (COST 231) and their maximum traffic load has been found for an operator with 36 TCH frequencies (9.8 MHz), ideal power control, and DTX with voice activity factor of 50%. The maximum capacity per site was obtained for a sectorised base station and a frequency reuse factor of 1/3 with a traffic load of 30% (DTX lowers the occupation of the traffic channels to approx. 15% of the time)3. The capacity increase of a random SFH network was approx. 74 % compared to a non SFH network with a frequency reuse scheme of 4/12. For omni-directional configurations, a 3 reuse scheme offered the greatest capacity increase. In general the capacity increase depends on the spectrum allocation: the higher the allocation (i.e. the number of hopping frequencies), the higher the increase. To take advantage of these capacity increases, a new approach for network planning is required. More bandwidth has to be assigned to each base station than is strictly needed if only its traffic load were taken into account. This approach is called fractional loading. In practise, use of the 1/3 scheme would make frequency planning trivial. Fractional loading can be obtained by either installing fewer transceivers than allocated frequencies and using synthesiser frequency hopping, or implementing an admission control procedure.

Here it must be reminded again that these results were obtained with homogeneous propagation and traffic conditions. The irregular propagation conditions and non homogeneous traffic distributions expected in real networks will allow higher fractional loading.
3

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 27 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

The need of admission control procedures is due to the fact that if as many TRXs as frequencies are installed, in the case of overload the network will not react by blocking calls, but by making all calls have a lower quality. This can result in more dropped calls. The effect of a dropped call for a subscriber is worse than that of a blocked call, and therefore they should be avoided. The advantage of this latter alternative is that local traffic peaks can be handled. A large number of channels are temporarily available in a sector, provided that the load in surrounding co-channel sectors is low, and the admission control procedures could make use of them. However, suitable algorithms are yet to be found. Until then fractional loading should only be used in conjunction with synthesiser hopping.

4.3.2.5 Multiple reuse patterns


The basic principle of multiple reuse patterns (MRP) is to divide the spectrum into subbands, each containing a different number of separately planned carriers. As a consequence, a network may operate with several reuse patterns simultaneously. One frequency from each sub-band is allocated to each sector, e.g. a 12 reuse for the BCCH, and 9 and 3 reuse for the TCH, respectively (referred to as 12/9/3). The total average reuse is then (12+9+3)/3 =8. Another possible plan is the 12/9/6/4 (reuse 7.75).
SPECTRUM ALLOCATION
BCCH-4/12 TCH1-3/9 TCH2-1/3

...

...

Figure 10 12/9/3 multiple reuse pattern The interference may vary greatly between different frequencies due to the difference in reuse. However, the interference diversity resulting from frequency hopping ensures that high quality is still maintained for all users, in spite of the tight reuse on some frequencies. Because the variation in interference is already achieved by the difference in reuse, this technique can be used with both cyclic and random hopping. The choice will generally depend on the number of transceivers in the cell. For few frequencies (for example, 2) cyclic is better because the spectrum utilisation will be better. A valuable feature of MRP is the ability to handle unevenly distributed traffic, i.e. different number of transceivers per cell. In the example given, one may not need a third transceiver in all cells initially, which means that the effective reuse on the third sub-band will be sparser than 3. As the capacity need increases, a third transceiver is installed in more cells, which results in gradual tightening of the average reuse. It is claimed that MRP can achieve a further capacity increase compared with the 3/9 reuse pattern (lower average reuse factors), and it offers the advantage of being suitable for baseband hopping, something that cannot be said of the 1/3 reuse pattern with fractional loading option. However, no studies have been done to find the most suitable MRP configuration in terms of capacity increase while maintaining good quality. Author : Date: RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 28 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

A 12/8/6/4 frequency plan has been used in a commercial GSM network, which roughly doubles the capacity compared to a standard 12 reuse pattern. The high system quality was shown to be maintained. Neither DTX nor power control were needed, because the downlink interference variation was achieved by the different reuse patterns. In another trial, an initial reuse factor of 16.9 requiring 40 carriers was tightened to a 14/10/6/2 configuration (reuse 12.87 and 32 carriers) and even a 12/10/4/2 (reuse 11.26 and 28 carriers). The BCCH carrier was non-hopping and no DTX or dynamic power control was used. There was no change in the minutes/dropped call. Some degradation of perceived speech quality in the second case was found, but it was shown to relate to interference in the tighter BCCH band. The number of successful handovers increased and in general they went more smoothly, showing that the new configurations had little impact on the quality. Simple calculations can be used to assess the capacity gains of these new configurations: Without FH: reuse 16.9, 40 channels => 2.48 TRX/cell, 12.23 Erlangs/cell. Phase 1: reuse 12.87, 40 channels => 3.1 TRX/cell, 16.7 Erlangs/cell (37% gain). Phase 2: reuse 11.29, 40 channels => 3.54 TRX/cell, 19.3 Erlangs/cell (58% gain).

4.3.2.6 Multiple reuse patterns and fractional loading


Multiple reuse patterns can be deployed in conjunction with fractional loading to further reduce the average reuse factor in a network and increase its capacity, without having to jump to the aggressive 1/3 pattern. It also allows the operator to plan for the future. The reuse strategy can be set tight to cope for future traffic increases. TRXs will be added to the sites when they are needed without changing the frequency plan. At the beginning the low fractional load will ensure high quality, which will then be traded off for capacity, when the latter is needed. One example of such configurations is a 12/8/5/4 with 40 carriers and a reuse of 6.2. The number of hopping frequencies is 5 or 6 (reuses 8, 5, 5, 4, 4, and 4 every other cell). The BCCH carrier is non hopping. An average of 4.0 TRX per cell would mean a 61% load and 21.4 Erlangs per cell. With 4.5 TRX (68% load) the traffic would increase to 24 Erlangs, and with 5.4 (82% load) to 30 Erlangs. The following table shows other configurations that have been proposed for a 40 carrier operator:

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 29 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

FL-MRP construction 12/8/7/4 12/9/7/4 12/8/7/5/4 12/9/6/3 12/8/7/3 12/9/6/3

Frequencies per cell 5.5 5.7 6.0 6.7 7.1 7.2

Average reuse 7.3 7.0 6.7 6.0 5.6 5.6

As it happens with MRP, no studies have yet been carried in order to find the most suitable configurations.

4.3.2.7 Fractional reuse patterns4


A different approach to tightening the frequency reuse are the fractional reuse patterns. They are obtained by dividing the set of available frequencies into overlapping subsets which are then assigned to the different cells that form a cluster. For example an operator with twelve frequencies and a 4/12 reuse structure, which normally would assign one frequency to each cell, can instead assign three frequencies with a fractional reuse pattern. This configuration is not reported to have been used in real networks because of its planning difficulty. It can be considered as a version of multiple reuse patterns with a non regular reuse in each frequency. A1 B1 C1 A2 B2 C2 A3 B3 C3 A4 B4 C4 f1 f2 f3 f4 f5 f6 f7 f8 f9 f10 f11 f12

4.3.2.8 Concentric cells


Concentric cells are used to increase capacity by enabling an operator to employ different frequency reuse patterns for different sets of frequencies allocated to a cell. Part of the spectrum is allocated to an outer zone for seamless contiguous coverage. This is planned using a conventional reuse patter. The remaining spectrum is divided into concentric zones, employing progressively tighter reuse patterns. Combining this structure with frequency hopping permits the network capacity to be enhanced even further by allowing a more aggressive frequency reuse in the different layers.

4.3.2.9 Control and traffic channels

Sometimes the term fractional patterns is used to refer to a tight reuse pattern with fractional loading. This is not the case in this document.
4

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 30 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Frequency hopping used in conjunction with power control and discontinuous transmission creates a potential for higher capacity in the system. Unfortunately, these features cannot be applied to frequencies used for broadcast control channels (BCCH frequencies). Furthermore, the interference on the BCCH frequencies should be kept low to ensure that cell selection, locating, access, paging, and so forth work properly. Today a 4/12 reuse factor is probably the best that can be achieved for the BCCH In many running systems, factor of around 15 is typically used. Consequently, to make the most of these techniques, it is desirable to have different reuse patterns for TCH and BCCH frequencies. Two different strategies can be envisaged: in the first one the frequencies are divided into two dedicated sub-bands: one for BCCH frequencies and another one for TCH frequencies. In the second BCCH and TCH frequencies share the same spectrum, although the BCCH are given a sparser reuse. Simulations show a much better performance of the dedicated bands strategy, achieved by limiting interference generated by BCCH frequencies to a specific band in the downlink, which is generally the limiting link in interference limited systems (diversity can be used in the uplink). It also ensures secure control channel behaviour independent of the traffic load. This better performance has also been observed in field measurements, and is therefore commonly used. It also has the advantage of avoiding most of the current handset problems, which arise when the control channels (which use the BCCH carrier) hop.

4.3.2.10

Effects on planning

Two are the major reasons that should encourage an operator to use frequency hopping in the traffic carriers and take advantage of its interference diversity effects: Increase quality in a network with interference problems. Increase the capacity of an already saturated network.

Increasing the quality


Just by switching on random frequency hopping among the traffic carriers in a network, with its existing frequency plan, a quality increase in terms of interference should be expected, unless of course the quality of the network is already good enough. The following paragraphs present the performance results from the activation of random frequency hopping 5 on an existing network. A significant improvement in the dropped call rate was seen both in the cells where frequency hopping was enabled and in the surrounding non-hopping cells. RXLEV/RXQUAL statistics were also collected. An increase in RXQUAL levels (i.e. a decrease in the measured quality) were reported. However, the average FER improved compared with the non-hopping case, i.e. there was a quality increase. This is because the channel decoder was able to correct errored bits on bad frequencies, using correct bits on other frequencies The increased RXQUAL levels produce an increase in the number of handover per call, which was an unwanted effect. In order to cope with it, network operators will need to alter RXQUAL-based handover thresholds in hopping cells. For example, the lower RXQUAL handover threshold can be increased by an amount equal to the increase in average RXQUAL values.
5

Baseband FH was used because filter combiners were used in all of the networks BTSs. RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 31 of 40

Author : Date:

Capacity Enhancement Core Techniques for GSM 900 and 1800

These results prove that SFH can be used to increase the QoS of an existing network. Nevertheless, frequency hopping for quality increase should not be used to hide major planning or tuning problems, in a badly planned network, i.e. with high and extended interference levels, because the performance would deteriorate.

Increasing the capacity


This section includes two cases: 1. An operator needs to introduce more TRXs for traffic reasons, 2. Some frequencies in the existing spectrum need to be freed for other layers. It had been seen in the previous sections that there are four ways to increase the capacity (efficiency) of the network by using random frequency hopping: tightening the reuse factor of the traffic carriers to 3/9, with downlink PC and DTX; using multiple reuse patterns; using multiple reuse patterns with fractional loading; tightening the reuse factor of the traffic carriers to 1/3 with fractional loading, downlink PC and DTX;

In reality they can all be considered steps in the process of tightening the average reuse factor in the network while maintaining sufficient quality. The latter two give the greatest capacity increase and flexibility, but they require synthesiser hopping. Therefore, they cannot be used in cells that have filter combiners, more than 4-6 TRXs (see section 4.4.2), or coverage restrictions that do not allow for the losses a hybrid combiner would introduce, i.e. cells where only baseband hopping is available. For networks with this type of cells the process of increasing capacity should be done in steps, gradually tightening the frequency reuse as more TRXs are needed in the network, and continuously monitoring the QoS attained. If homogeneous reuse patterns are used, downlink power control and discontinuous transmission will probably be needed. Multiple reuse patterns offer in this case greater flexibility, in that they are suited for non homogeneous traffic conditions, and non whole number average reuse factors can be achieved. For example, a network can go from a 12, to a 12/9 (10.5 average reuse), a 12/9/6 (9 average reuse) and a 12/9/6/3 (7.5) reuse configuration as the traffic increases. With synthesiser hopping, two different strategies to increase the capacity of the network can be followed. The first one is the same one used with baseband hopping: progressively tighten the reuse factor as more capacity is needed in the network. The only difference in this case is that fractional loading techniques are available to further decrease the average reuse factor. It must be ensured, however, that the appropriate fractional load is achieved. The second strategy was already outlined in section 4.3.2.6. The average reuse factor could be initially set tight, to cope for future traffic increases, and TRXs added to the sites when they are needed, without changing the frequency plan. The reuse strategy can be either a 1/3 or an MRP. As new TRXs are added, the QoS should be monitored to ensure that the maximum fractional load the plan allows for is not surpassed. DTX and PC can be deployed to increase this maximum fractional load. The choice between 1/3 and MRP is still difficult because of the lack of comparison studies. Both of them are suited to non-homogeneous traffic conditions. The advantage of the 1/3 reuse pattern is that it eliminates the need for frequency planning of the traffic carriers in a network, and therefore Author : Date: RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 32 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

it is very flexible when it comes to introducing new base stations. However, it is also a very aggressive pattern. A possible use of 1/3 patterns is the microcell layer. In general, reuse patterns are suitable for networks with a regular layout. Most of the time this is not the case because of the irregular terrain morphology and propagation conditions, and uneven traffic distributions. This is why frequency planning is often done with the aid of automatic planning tools which base their frequency assignments on interference matrices. Such an approach, which is based on a certain model of the network, is however not suitable for planning frequency hopping. New modelling techniques are required in the existing automatic planning tools in order to make the most of frequency hopping. Most operators already have a planning tool at their disposal. With it, frequency hopping planning can be tackled in different ways, depending on the chosen strategy. Fractional loading can be achieved by assigning to each cell more frequencies than the traffic requires, i.e. assigning more TRXs in the frequency planning tool, even though they will not be real TRXs. An homogeneous reuse pattern can be tightened by reducing the required C/I planning threshold. The difficulty is to determine the level to which the C/I threshold can be reduced. A possible approach is to decrease its value in steps, monitoring the network performance at each step. Multiple reuse pattern can be planned by following a step approach. First, the first layer of TRXs (which represent frequencies in the tool) is planned using the conventional C/I threshold. In the second step, the first layer of frequencies remains fixed, and an additional layer of TRXs (frequencies) is added to the base stations that require them. Frequencies are assigned using a new, lower C/I threshold. In the third step, a new layer of TRXs (frequencies) is introduced and the same procedure is followed. Steps are repeated until all the layers have been assigned.

4.4

Lucents solutions

4.4.1 Base station equipment


RBS900 family : 900 band - will only support baseband hopping. BTS2000 family : 900/1800 bands. Bosch RFUs: The second letter in their code - B => baseband hopping, S => baseband/synthesiser hopping. Lucents SRFUs are capable of synthesiser and baseband hopping. CUBE supports synthesiser hopping.

4.4.2 Antenna coupling equipment


Filter configurations (with TXFU09/TXFU18 filter combiners) only allow baseband frequency hopping. Hybrid and diplexer configurations (which either use hybrid combiners - TXDU09/TXDU18) allow the two types.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 33 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Due to the hybrid combiner losses (each layer of them introduces a loss of 3 dB into the overall combining loss), hybrid configurations of up to 4 TRX are available with two antennas and 6 TRX with three.

4.4.3

Fill-sender and phantom RTs

The BCCH carrier is always transmitting to enable mobiles to monitor it. This does not cause a problem for baseband hopping since there are as many RFUs as frequencies, and one of them is available to transmit the BCCH carrier. For synthesiser hopping, the traffic channels assigned to the carrier which includes the BCCH, will not be able to hop unless an additional transmitter is used. This additional transmitter is known as the fill-sender, but would be more accurately termed a fill transmitter. It is a RFU or SRFU which is used to transmit continuously on the BCCH frequency. To implement hopping channels on the BCCH-RT the fill-sender RFU or SRFU must be located in the same DRCC (Double Radio Codec and Control 6) as the BCCH-RT, in the BTS. Fillsenders cannot, therefore, be used on the 6od or the BTS2000/2C as these do not have the DRCC as part of their physical configuration.

DRCC

SRFU0 (synth hopping)

Fill Sender

SRFU1 (synth hopping)

F1 F2 F3

F0

F0

F1 F2 F3

FCO

CCB
Figure 11 BTS with fill-sender configuration

CCB

FCO

In the case of baseband hopping, there also exists the possibility of hopping over more frequencies than RTs, by adding phantom RTs. Like the fill-sender, this is an extra RFU set on a different frequency, that can be included in the hopping sequences, but does not carry additional traffic. Both the fill-sender and the phantom RT take up the physical space of a standard transmitter, and their suitability has to be weighed against the fact that the equipment could be used to support another 8 traffic channels.

4.4.4 Possible hopping configurations 4.4.4.1 Limitations


It is a logical association of two slots which are always the ones that are powered by the same power supply unit.
6

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 34 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

Frequency hopping in a cell is based on the concept of a Frequency Hopping System (FHS) which consists of a set of frequencies (from the pool of frequencies that are available at the cell), and a Hopping Sequence Number (HSN). As seen in 4.2.2, the HSN is used to generate the order in which the frequencies will be used. A value of 0 specifies a cyclic hopping sequence. All of the other values (1-63) cause a quasi-random sequence to be generated. The frequencies in a FHS must obey the co-site minimum spacing rule: there must be a separation of at least three GSM carriers between them. Each channel, defined by a TRX and a time-slot, can be assigned to a FHS and given a Mobile Allocation Index Offset (MAIO), provided that the assignment does not cause an Um interface collision (channels using the same frequency at the same time). Because the frequency redefinition procedure has to be completed in about 3.5 minutes, the configuration must comply with the following limitations: Maximum number of frequencies in a FHS: 8. Maximum number of FHS using the same frequency: 2. Maximum number of physical channels using the same FHS: 42. Maximum number of FHS in a BTS: 8. Maximum number of FHS in a BSS: 48. These limits impose important constraints, because they limit the number of hopping frequencies in a cell to 8 in the case of synthesiser hopping. This restricts the possibility to exploit the capacity gain of fractional loading. Means to change these limits are being studied. However, it still remains an open issue. Frequency hopping is allowed with concentric cells, as long as hopping is between frequencies assigned to the same zone. For dual band operation, frequency hopping is only allowed between frequencies belonging to the same band.

4.4.4.2 Recommended configurations


In general it is recommended to include the given RTs and channels in a frequency hopping system in such a way that the number of frequencies are maximised per FHS. According to the limitations described above (which remain an open issue) and the restrictions imposed by the type of frequency hopping systems, the following configurations are recommended:

RBS900
Number of RTs (= n) n=2 2 < n <= 6 6 < n <= 10 Number of FHS 1 2 3 Time-slots / No. of TRXs per FHS FHS:0 1..7 / n 1..7 / n 1..7 / (n/2) 2) FHS:1 -/0 / (n-1) 1) 0 / (n/2-1) 2) FHS:2 -/-/0..7 / (n-n/2) 2)

1) frequency of BCCH-RT excluded 2) where n/2 is rounded up in case of n odd

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 35 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

BTS2000
Number of RTs Number of FHS (= n) n=2 1 2 < n <= 6 2 6 < n <= 10 3 10 < n <= 12 4 Time-slots / No. of TRXs per FHS FHS:0 1..7 / n 1..7 / n 1..7 / (n/2) 2) 1..7 / (n/2) 2) FHS:1 -/0 / (n-1) 1) 0 / (n/2-1) 2) 0 / (n/2-1) 2) FHS:2 -/-/0..7 / (n-n/2) 2) 0..3 / (n-n/2) 2) FHS:3 -/-/-/4..7 / (n-n/2) 2)

1) frequency of BCCH-RT excluded 2) where n/2 is rounded up in case of n odd

4.4.4.3 Mixed operation of hopping and non-hopping modes


This is allowed as long as it does not cause collisions at the Um interface. The following rules have to be taken into account: the frequencies of a hopping system associated with a channel (TRX and time-slot), must not be used in other hopping systems associated with a channel using the same time-slot; a non hopping channel must not be assigned a frequency that is used in a hopping system with associated channels using the same time-slot.

4.4.5 Feature activation and parameters


Once the hopping configuration has been decided, frequency hopping has to be configured and activated in the elements involved. 1. BTS hopping mode : The hopping mode (baseband or synthesiser) is implicitly defined by the BTS-HW configuration for RBS900. In the case of the BTS2000 family, which allow both types of hopping, it can be changed via the RBT20007 (Radio Base station Tester). If a fill-sender is used, the corresponding RT should be configured via the RBT2000. 2. BSS feature enabling : LM3.0 software release only allows baseband hopping, for synthesiser hopping LM4.0 (GSM release 7.0) should be used. The new software, if not installed can be downloaded via the OMC. Frequency hopping is a purchased option, and a special code is required from the factory to allow it to be enabled by site specific setting the variable Freq_Hopp_Enabled in the BSS local configuration data (lcd). 3. OMC feature configuration: The activation of frequency hopping requires the creation and modification of different objects on a per cell basis8: All the frequencies of the hopping systems have to be present in the attribute cellAllocation (CellAllocation,CELLALLOC) in the BTS object. In the case of baseband hopping, for every frequency to be used in a frequency hopping system, a Radio Terminal (RT) has to be created in advance. The frequency has to be defined by the attribute initialFrequency (InitialFrequency,INITFREQ). The creation of a phantom-RT is possible by dropping the following attributes: AbisServiceProvider , AbisSigHDLCInfo, AbisTrafSlotInfo , BackupObject.

7 8

It is a PC (notebook) with special software used to administrate the BTS2000. Internal parameter names are given. OMC GUI and AUI parameter names follow in brackets. RF Systems & Capacity Group 9th April 98 Issue: Page: 1.0 36 of 40

Author : Date:

Capacity Enhancement Core Techniques for GSM 900 and 1800

One FH object must be created for each of the hopping sequences to be used, with object attributes: allocatedFrequencies (AllocatedFrequencies, ALLOCFREQ), which must all be present in the cellAllocation attribute of the BTS. HSN (SequenceNumber, HOPSEQNO), which defines the Hopping Sequence Number 0 in case of cyclic hopping. a unique number 1-63 in different cells, which have the same group of frequencies, in random hopping. Each frequency in allocatedFrequencies must not be associated with more than one other hopping system. Up to eight FH objects may be created per BTS. The Channel CHN objects must be created as hopping or non hopping by appropriate setting its attributes: freqHoppRelationship (FHRelationship, FREQHOPREL), which either points to an associated object FH or is set to non-hopping. channelType (ChannelType)must not be a CCCH if it is set to hopping. The RT that the channel belongs to must not be a phantom-RT. A MAIO will be generated internally for the channel according to GSM Rec. 05.02. The number of channels associated with the same hopping system must be 42 at the most. The setting of these parameters can be easily done using OMC feature omc-cm093: Automatic Network Modification for Frequency Hopping, which gives the frequency planner the ability to provide the OMC2000 operator with frequency/frequency hopping plans in an electronic file. The OMC reads the plan, validates the data, and generates a set of AUI commands that convert the OMC data into the data found in the plan. The generated AUI scripts may be executed immediately or scheduled for later execution. Once the frequency hopping systems have been activated, a reconfiguration process takes place which consists of two different procedures: The reallocation procedure, where the BTS is provided with the necessary information, and is ordered to reconfigure its hopping behaviour at a given starting time. The frequency redefinition procedure, which triggers the call handling function to start the frequency redefinition (see 4.2.5).

4.4.6 Fault defence mechanisms


Frequency hopping is autonomously disabled in the following situations: Baseband hopping: The number of available frequencies used by the hopping system falls below a given threshold due to severe RT faults. The threshold is calculated by multiplying the number of frequencies by a percentage defined in the local configuration area of the BCE (LMB Enable/Disable CH options). Synthesiser hopping: Fill-sender failure if installed.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 37 of 40

Capacity Enhancement Core Techniques for GSM 900 and 1800

5.

CONCLUSION
The use of frequency hopping can provide the operator with two benefits: An increase in the quality of both signal level and interference for slow moving mobiles. A decrease in the effect of interference when used with DTX and power control.

The first enables the operator to plan the frequency allocations with the C/I thresholds of medium speed mobiles, instead of the worse case slow mobiles. The second allows the operator to plan with smaller C/I threshold values. Consequently, when properly used, frequency hopping can help the operator increase the capacity of their network beyond the limits imposed by their spectrum allocation, especially when combined with discontinuous transmission and power control. When compared when other techniques of capacity enhancement, like concentric cells, hierarchical structures or dual band, frequency hopping proves to be the least expensive one, both in terms of cost and effect on network performance (for example, number of handovers). All these makes it a very attractive technique to start growing the capacity of the network. The gains of frequency hopping when used for interference diversity depend on the number of hopping frequencies. It is, therefore, important that the limits that Lucents equipment impose on that number are reviewed, so that Lucent can offer their customers the best possible configuration to allow them to take advantage of frequency hopping. The process has already been started.

Author : Date:

RF Systems & Capacity Group 9th April 98

Issue: Page:

1.0 38 of 40

You might also like