Professional Documents
Culture Documents
Previous lectures
Fixed routing infrastructure Fixed infrastructure provides routing RIP, OSPF were designed for fixed routing infrastructure
Routers are in fixed infrastructure Routers are not mobile
CSE, UNSW
This lecture
What if there is no routing infrastructure? Ad hoc networks
What are they? Routing in MANETs Future directions
CSE, UNSW
CSE, UNSW
CSE, UNSW
C A B D
6
CSE, UNSW
Battle field communications e.g. among tanks and soldiers Military origin of MANET CSE, UNSW
7
Search and rescue Mobile devices on personnel and vehicles CSE, UNSW Good for remote areas
CSE, UNSW
http://yoshida.kuee.kyoto-u.ac.jp/~aolim/text/adhoc/text/8_1.html 9
Why MANETs
Why MANETs?
Fast and easy to deploy Wireless untethered connection Mobility is assumed Anywhere anytime network formation
CSE, UNSW
10
Features of MANET
Nodes use wireless communications Lack of fixed infrastructure Multi-hop radio communications Nodes are also routers Variable topology
Nodes can only directly communicate with nodes within their radio range [illustration - next slide] Nodes forward the traffic for each other In contrast to the Internet: Hosts do not do routing
CSE, UNSW
11
C A
Wireless connectivites A B, B C, D B, C D
B D
12
A-C : multi-hop
CSE, UNSW
13
In MANETs
topology may change rapidly no fixed neighbours (neighbours change dynamically) no fixed routes (they may quickly become invalid)
CSE, UNSW
14
Later
B A C B
Topology
Topology
CSE, UNSW
15
wireless bandwidth is limited and expensive routing overhead (bandwidth consumption) has to be very low routers are not connected to fixed powerlines routing overhead (power consumption) has to be low
CSE, UNSW
16
Topology needs to be updated to provide accurate topology information Accurate route information But frequent topology update consumes bandwidth
CSE, UNSW
17
CSE, UNSW
18
CSE, UNSW
19
Approach 4: Position-based
Packet header carries the position (e.g. GPS coordinates) of the destination node Routing decision is based on position rather than IP addresses
CSE, UNSW
20
CSE, UNSW
21
CSE, UNSW
22
Components
CSE, UNSW
23
AODV - terminology
Originating node
The node which initiates the route discovery process
CSE, UNSW
24
CSE, UNSW
25
CSE, UNSW
26
When a node receives an RREQ message, it enters reverse path information in its routing table
CSE, UNSW
RR E
RR E
5
RREQ
RR E
RREQ
3 5
Later, Node 5 receives the same RREQ from node 3, should it modify its routing table?
28
CSE, UNSW
Exercise
1 2 3 5 6 9 D
Assuming Node 1 = Source, Node D = Destination. If no intermediate nodes knows a route to D. What are the reverse paths that have been established when the RREQ message reaches the destination?
CSE, UNSW
29
Solution
1 2 3 5 6 9 D
Assuming node-3 receives RREQ from node-2 before it receives the request from node 5, and node-D receives it from node-6 before from node-9.
CSE, UNSW
30
An RREQ message is uniquely identified by source address and request id The message is broadcast by using IP limited broadcast address 255.255.255.255 Each intermediate node increments the hop count field by 1
CSE, UNSW
31
6
CSE, UNSW
D 9
= RREQ
= RREP
D unicasts an RREP message to the node from which it first receives an RREQ message
32
CSE, UNSW
33
1 2 3 5 6 9 D
When Node 6 receives the RREP message 1. It enters the following information in its routing table Destination = D, NextHop = D. 2. It unicasts the RREP message along the reverse path that has already been established.
Exercise:
= reverse path established = RREP
1 2 3 5 6 9 D
When Node 3 receives the RREP message from node 6, what information will it enter in its routing table? What actions will it take after that?
Solution:
= reverse path established = RREP
1 2 3 5 6 9 D
Since it receives the RREP message from node 6, it enters Destination = Node D, Next Hop = Node 6. It then consults its routing table and find that the next hop to Node 1 (the originating node) is Node 2 (this is the reverse path) so it unicasts an RREP to Node 2
1 2 3 5
CSE, UNSW
= RREQ
= RREP
6 9
Route Maintenance
Due to movements of nodes, routes may be broken during a session Such broken routes must be repaired and maintained until the end of the session The node which detects a link-break in an active route, back propagates an error message to the source The source reinitiates the route discovery process to find a new route
CSE, UNSW
38
4
error mesg
2 1
4 6 5
2 1
CSE, UNSW
3 6
5
39
CSE, UNSW
40
CSE, UNSW
41
1
CSE, UNSW
4 6 5
Only shaded entries are affected when 4 becomes unreachable to 3. Active 42 neighbour 2 has to be notified.
CSE, UNSW
43
CSE, UNSW
44
CSE, UNSW
45
CSE, UNSW
46
More route discovery rounds are required if the originating node still fails to find a route
For each new round, the TTL is incremented by TTL_Increment This is continued until TTL reaches TTL_Threshold Initiate network-wide broadcast by setting TTL = Net_Diameter
CSE, UNSW
47
ERS parameters
The RFC recommends the following setting
TTL_START = 1 TTL_Increment = 2 TTL_Threshold = 7 Net_Diameter = 35 Node_Traversal_Time = 40ms Ring_Traversal_Time = 2 x Node_Traversal_Time x (TTL_value + Timeout_Buffer) Timeout_Buffer = 2
CSE, UNSW
48
Sets TTL = 1
2 1
Search area
CSE, UNSW
4 6 5
7 8
Originating node timeout as no RREP is received. 49 It sets TTL = 3 in the next round of search.
2 1
Search area
4 6 5
7 8
CSE, UNSW
Note that RREQ does not reach nodes 5,7 and 8. Thus saving them from processing RREQ messages.
50
CSE, UNSW
51
CSE, UNSW
52
References
RFC3561 Ad hoc on-demand distance vector (AODV) routing
CSE, UNSW
53