ABSTRACT: Video Dissemination over Hybrid Cellular and Ad Hoc Networks

We study the problem of disseminating videos to mobile users by using a hybrid cellular and ad
hoc network. In particular, we formulate the problem of optimally choosing the mobile devices
that will serve as gateways from the cellular to the ad hoc network, the ad hoc routes from the
gateways to individual devices, and the layers to deliver on these ad hoc routes. We develop a
Mixed Integer Linear Program MILP!"based algorithm, called P#P$, to solve this optimi%ation
problem. Pocket delivers the highest possible video &uality and optimi%ation problem that
determines'
(! $he mobile devices that will serve as gateways and relay video data from the cellular network
to the ad hoc network,
)! $he multihop ad hoc routes for disseminating video data
*! $he subsets of video data each mobile device relays to the next hops under capacity
constraints. We formulate the optimi%ation problem into a Mixed Integer Linear Program
MILP!, and propose an MILP"based algorithm, called P#P$, to optimally solve the problem.
We recommend the $+, algorithm for video streaming over hybrid cellular and ad hoc networks.
Last, we also build a real video dissemination system among multiple -ndroid smart phones over
a live cellular network. .ia actual experiments, we demonstrate the practicality and efficiency of
the proposed $+, algorithm. We call it $ree"/ased +euristic ,cheduling $+,! algorithm, and it
works as follows' We first sort all the transmission units in the W"segment scheduling window in
descending order of importance, by layer, segment, and video. We then go through these WL
units, and se&uentially schedule the transmissions to all mobile devices.
E!ST!N" S#STE$:
Linear Program LP!"based algorithm called M$,, for lower time complexity generic ad hoc
protocols do not work well in hybrid cellular and Wi0i ad hoc networks, and may lead to'
(! degraded overall throughput, )! unfair resource allocation, and *! low resilience to mobility.
$hey propose two approaches to improve the efficiency of ad hoc protocols. 0irst, the base
station can run optimi%ation algorithms for the Wi0i ad hoc network, for example, to build
optimi%ed routes. ,econd, mobile devices connected to other access networks can offload traffic
from the cellular network to those access networks, so as to avoid network congestion around the
base station.
D!SADVANTA"ES:
1xisting algorithms achieve at least (2 d/ &uality improvements and result in up to 324 packet
delivery delay reduction.
%R&%&SED S#STE$:
We propose a hybrid network, in which each multicast group is either in the cellular in the ad hoc
mode. Initially, all multicast groups are in ad hoc mode, and when the bandwidth re&uirement of
a group exceeds the ad hoc network capacity, the base station picks up that group and switches it
into the cellular mode. In the ad hoc network, a flooding routing protocol is used to discover
neighbors and a heuristic is employed to forward video data. #ur work differs from in several
aspects' (! we propose a unified optimi%ation problem that 5ointly finds the optimal gateway
mobile devices, ad hoc routes, and video adaptation, )! we consider existing cellular base
stations that may not natively support multicast, and *! we employ .ariable"/it"6ate ./6!
streams.
More specifically, we empirically measure the mapping between the node location and link
capacity several times, and use the resulting values for capacity estimation. We adopt the video
traces of +.)789MP1:8 layered videos from an online video library. $he mean bit rate and
average video &uality for each layer of the considered videos are given in $able ). In this paper,
we report sample simulation results of distributing ;rew. +owever, the proposed formulation and
solutions are general and also work for the scenarios where mobile devices watch different
videos.
ADVANTA"ES:
(. $he links into mobile devices on breadth"first trees of transmission units with higher &uality
improvement values are given higher priorities.
). $he links with higher ad hoc link capacities are given higher priorities.
*. $he links from mobile devices with higher cellular link capacities are given higher priorities.
$&D'(E DESCR!%T!&N:
SERVER C(!ENT $&D'(E:
;lient"server computing or networking is a distributed application architecture that partitions
tasks or workloads between service providers servers! and service re&uesters, called clients.
#ften clients and servers operate over a computer network on separate hardware. - server
machine is a high"performance host that is running one or more server programs which share its
resources with clients. - client also shares any of its resources< ;lients therefore initiate
communication sessions with servers which await listen to! incoming re&uests.
)!$A RE(A# NET)&R*S:
WiM-= bandwidth allocation schemes in employ multiple loops to examine the performance of
the different combinations of recipients, which results in extremely high computational
complexity. $he bandwidth allocation scheme proposed in this study applies greedy methods to
achieve low computational complexity while incorporating the table"consulting mechanisms to
avoid redundant bandwidth allocation scheme can efficiently allocate bandwidth while
maintaining low computational complexity. WiM-= provide diverse data rates, +.)789,.;
allow a video stream to be split into one base layer and multiple enhancement layers. $his study
assumes that a video can be split into six layers one base layer and five enhancement layers!
corresponding to the six video &uality levels a user with the re&uirements of 78kbit9s ()3 kbit9s
can be satisfied by receiving the base layer and one enhancement layer.
RES&'RCE A((&CAT!&N:
#ur resource allocation model for two"hop WiM-= relay networks consists of one /,, M 6,s,
and > ,,s. 0or consistency, the /, is regarded as the 2th 6, and is denoted by 6,2 in the
following discussion, while the 6,s are denoted by 6,( to 6,M.-n ,, can associate either with
the /, or with one of the 6,s, and the number of ,,s associated with 6,m is denoted by >m.
$he notation ,,m<n represents the nth ,, associated with 6,m.
;?m represents the channel &uality of the link between the /, and 6,m while ;?m<n
represents the channel &uality between 6,m and ,,m<n. -ssume that the video streams for the
links with lower channel &uality should be transmitted by the modulation schemes with higher
reliability.
V!DE& STREA$!N":
,calable video broadcast9multicast solutions efficiently integrates scalable video coding, *:
broadcast and ad"hoc forwarding to balance the system"wide and video &uality of all viewers at
*: cell. In our solution, video is downloading into multiple layers. $he base station broadcasts
different layers at different rates to cover viewers at different ranges. -ll viewers are guaranteed
to receive the base layer, and viewers closer to the base station can receive more enhancement
layers. @sing WiM-= 6elay >etworks connections, viewers far away from the base station can
obtain from their neighbors closer to the base station the enhancement layers that they cannot
receive directly from the base station. #ur solution strikes a good balance between the average
and worst"case performance for all viewers in the cell. We design multi"hop relay routing
schemes to exploit the broadcast nature of ad"hoc transmissions and eliminate redundant video
relays from helpers to their receivers.
+'A(!T# &%T!$!,AT!&N:
#ur channel &ualities of these links, /,s and 6,s can dynamically adapt the downlink
modulation and coding schemes M;,s! for data transmission. When 6,s are deployed at
appropriate locations between the /,s and ,,s, the end"to"end channel &ualities can be improved
and the /,s and 6,s can adopt high data"rate M;,s. /ased on this improvement in data rate,
I111 32).(75 systems can offer higher throughput and serve more users than I111 32).(7e
systems. /ased on the performance enhancements above, I111 32).(75 has the potential to
provide real"time video multicast services such as mobile IP$., live video streaming e.g.,
athletic events!, and online gaming!.
+owever, the /,s should allocate bandwidth efficiently to support such bandwidth"hungry
services while guaranteeing the &uality of user experience ?o1!. $he bandwidth allocation
problems in I111 32).(75 networks are more challenging than those in I111 32).(7e networks
because the /,s allocate bandwidth not only to the ,,s, but also to the 6,s. Multicasting also
complicates the bandwidth allocation problems of these factors, designing an efficient bandwidth
allocation scheme for video multicast services.
We have presented various bandwidth allocation approaches for video services in I111 32).(7e
networks i.e., single"hop WiM-= systems!. $he approaches in and allocate bandwidth by
exploiting the common technology of scalable video coding ,.;! specified in the +.)789,.;
standard. $he +.)789,.; standard is extended from +.)789-.;, and can further split a video
stream into a base layer for providing the basic video &uality and multiple enhancement layers
for providing better video &uality layer by"layer.