Jawarilal, Ajeeth
CS898T 

Topic: Integrating Multipoint Relay into SARP

Goals:
1. Bandwidth savings and reduced control overhead compared to OLSR by 
adopting a reactive approach to finding routes.
2. Optimized MPR broadcast solves the broadcast storm problem in Manets.
3. Latency in finding routes is less compared to DSR and AODV. 
4. Greater support for QoS because delay and throughput information can be 
maintained.

Background:
Optimized Link State Routing Protocol:
OLSR is a proactive link state routing protocol that optimizes link state 
broadcast using Multi Point Relays (MPR) [3]. All nodes maintain local 
connectivity by broadcasting Hello messages to their one hop neighbors. 
The Hello messages of a node contain information about nodes with which it 
has bi-directional and unidirectional link connectivity. Using this local 
connectivity information each node chooses its MPR set. The MPR set of a 
node is a set of one hop neighbors that gives it full two hop 
connectivity. 

To start with all the one hop neighbors of a node may be included in its 
MPR set. After this the MPR set is optimized and the smaller the MPR set, 
more optimized it is. A node that is a MPR maintains a Topology Control 
(TC) table which consists of all nodes that have chosen it as their MPR - 
"MPR Selectors". All MPRs broadcast the TC information proactively. All 
nodes receive this information and process it but only the MPRs 
rebroadcast it thereby optimizing the link state broadcast. Any change in 
the link state is reflected in the Hello message and updated in the TC 
table which is proactively broadcast. If any node needs a route to another 
node, it can be found by processing the information in the TC updates 
received.

Self learning Ad Hoc Routing Protocol(SARP): 
The routing process in reactive routing protocol like SARP consists of 
Route Discovery and Route maintenance [1]. When routes to a destination 
are not known, Route Discovery is performed by broadcasting RREQ using 
expanding ring approach. New routes are learnt by intermediate nodes by 
source routing the RREQ, RREP and RERR packets. The destination node or 
any intermediate node that has a fresh enough route to the destination 
replies with RREP. Route freshness is determined using sequence numbers. 
If any route expires due to node mobility a RERR message is broadcast to 
all nodes that use that node for their routes. 

Scope:
The proposal of integrating MPR into SARP provides us with a hybrid 
approach to solving the routing problem in Manets. In the local vicinity 
of a MPR the link sate information is maintained proactively but a route 
to a distant node is sought reactively. 

In the proposed routing protocol all nodes broadcasts Hello messages.
Hello messages to keep track of all neighbors with which it has 
bi-directional and unidirectional connectivity. Based on this information 
in hello messages each node will select its MPR set. Many nodes will have 
over lapping MPR set. The MPR nodes maintain a TC table which consists of 
all neighbors that have chosen it as their MPR - "MPR selectors". In OLSR 
this TC information is broadcast proactively. The proposed protocol adopts 
a reactive approach when sharing the TC information. 

Whenever a route to a node is needed the requesting node does a one hop 
RREQ broadcast to its MPR set. The MPRs will in turn rebroadcast this RREQ 
to other MPRs and eventually it is propagated to all the MPRs in the 
network. This MPR-only broadcast leads to a much optimized broadcasting 
technique when compared to simple flooding and thereby eliminates the 
Broadcast Storm problem in Manets. The MPRs that have a fresh enough route 
to the destination or the destination MPR replies with a RREP. The route 
selection can be done based on minimum number of hops to the destination. 
Since all RREQ, RREP, RERR packets are source routed in SARP, automatic 
learning of routes is performed thereby reducing the number of RREQ.

References:
1.Siddartha Gundeti, Tish Best, Ravi Pendse, "Self-Learning Ad-Hoc Routing
Protocol", VTC 2003 
2. Jatinder Pal Singh, Nicholas Bambos, Bhaskar Srinivasan and Detlef
Clawin, "Proposal and Demonstration of Link Connectivity Assessment
based Applications to Routing in Mobile Ad-hoc Networks", IEEE Vehicular Technology Conference, 2003, Orlando, Florida
3. T. Clausen, P. Jacquet, A. Laouiti, P. Muhlethaler, a. Qayyum, and
L. Viennot. Optimized Link State Routing Protocol. In IEEE INMIC,
Pakistan, 2001.
4. Sung-Ju Lee, Elizabeth M. Royer, and Charles E. Perkins.
"Scalability Study of the Ad hoc On-Demand Distance Vector Routing Protocol." 
International Journal on Network Management, 13(2),March-April 2003.