13-08-2012, 03:07 PM
Cognitive Multipath Multi-Channel Routing Protocol for Mobile Ad-Hoc Networks
Cognitive Multipath Multi-Channel.pdf (Size: 330.67 KB / Downloads: 26)
Abstract
This paper introduces a novel cognitive multipath
multi-channel routing protocol (CMMRP) for mobile ad-hoc
networks. It is designed for a multi-channel environment where
nodes can simultaneously use multiple interfaces to transmit
packets over different frequencies. It employs cognitive functions
to make nodes intelligently select multiple node-disjoint,
edge-disjoint and frequency-disjoint paths. Neural network
machine learning is adopted to make nodes aware of history.
CMMRP employs a modified path discovery protocol which can
be divided into two parts, a space discovery and spectrum
discovery. Simulation results show that CMMRP significantly
improves network reliability and performance.
Index Terms— multipath, multi-channel, cognitive, machine
learning, mobile ad-hoc network, prediction
I. 0BINTRODUCTION
Multipath routing protocols have been proposed in many
papers for both wired networks and wireless networks. In
wired networks, the major consideration is how nodes utilize
multiple paths such as back up redundant paths and load
balancing among available paths. Currently, OSPF and BGP-4
are the dominant routing protocols for wired networks. They
both include multipath capabilities. Based on desired metrics,
nodes are able to balance load among paths or support
redundant paths. On the other hand, mobile ad-hoc networks
(MANETs) have two new characteristics, wireless
communication and dynamic physical topologies. Nodes have
to suffer interference from other nodes because of the
broadcast nature of wireless communications. As a result,
network performance degrades if neighboring nodes transmit
packets over the same frequency. This problem becomes
serious for multipath routing protocols when nodes
simultaneously use multiple paths to transmit packets. Also,
dynamic physical topology usually incurs considerable
overhead to repair broken paths especially for multipath
routing protocols. Nodes should select stable links to save
routing overhead and maintain network reliability. On the
other hand, it is hard to perform optimal load balancing in
dynamic physical topologies because significant overhead has
to be generated to inform nodes of updated conditions for each
path. Considering the characteristics of MANETs, we argue
that multipath routing protocols on MANETs should be
focused on how nodes select multiple paths to improve
network reliability and performance instead of how nodes
optimally utilize multiple paths.
The problem we address in this paper is how nodes select
multiple node-disjoint, edge-disjoint, and frequency-disjoint
paths. Many multipath routing protocols have been proposed
to allow nodes to select multiple node-disjoint and
edge-disjoint paths. However, few of them provide solutions
for how nodes can select multiple frequency-disjoint paths.
Our main contribution is the cognitive multipath
multi-channel routing protocol (CMMRP). It falls into a novel
category of routing protocols, cognitive routing protocols.
Each node predicts future conditions of links and frequencies
based on past experience. Multiple disjoint paths are
discovered one by one by triggering RREQ packets multiple
times from the source node, trading routing overhead for
network reliability and performance.
This work is part of ongoing CogNet effort [1-4], which is
focusing on applying cognitive techniques to improve network
performance. The rest of paper is organized as follows. We
describe related work in Section II, discuss the approach in
depth in Section III, describe simulation results in Section IV,
and conclude in Section V.
II. 1BRELATED WORK
In this section, we discuss work related to CMMRP.
A. 5BDisjoint Multipath Routing Protocols
A multitude of disjoint multipath routing protocols have
been proposed (e.g. [5-10]). The main idea is that a source
node selects multiple non-overlapped paths to transmit packets
to a destination node. In [5], the authors proposed a maximally
disjoint multipath routing protocol, an on-demand routing
protocol. Unlike traditional protocols where intermediate
nodes discard duplicated RREQ packets, the proposed routing
protocol lets intermediate nodes relay duplicated RREQ
packets to make destination node know all possible paths. A
destination node selects multiple disjoint paths according to
recorded traversed paths of RREQ packets and sends RREP
packets back to source node through the corresponding paths.
Disjoint multipath routing protocols improve network
reliability because the possibility of multiple disjoint paths
breaking simultaneously is much lower than possibility that
one path breaks. However, the main drawback is the
significant routing overhead incurred when intermediate nodes
relay all duplicated RREQ packets.
Suyang Ju and Joseph B. Evans
{jusuyang,evans}[at]ittc.ku.edu
University of Kansas
Lawrence, Kansas, USA
Cognitive Multipath Multi-Channel
Routing Protocol for Mobile Ad-Hoc Networks
978-1-4244-5638-3/10/$26.00 ©2010 IEEE
This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE Globecom 2010 proceedings.
B. 6BMeshed Multipath Routing Protocols
Many meshed multipath routing protocols have been
proposed (e.g. [11-15]). The main idea is that a source node
selects multiple overlapped paths to transmit packets to a
destination node. Unlike disjoint multipath routing protocols,
they let intermediate nodes have multiple paths between
source node and destination node. As a result, there will be a
large number of overlapping paths between source node and
destination node constructed via links selected by intermediate
nodes. In [11], the authors argue that meshed multipath
routing protocols are more reliable than disjoint multipath
routing protocols. However, the difference is small because
the selected paths are overlapping but not independent.
Meshed multipath routing protocols improve network
reliability considerably because of the large number of
overlapping paths between source node and destination node.
However, the main drawback is the excessive overhead
incurred when intermediate nodes transmit replicated data
packets over multiple links to the destination node.
C. 7BCognitive Routing Protocols
In recent years, several cognitive routing protocols have
been proposed (e.g. [16-19]). The principle idea is that nodes
are able to make wise decisions by predicting future network
conditions based on past experience. Machine learning
techniques are adopted to make nodes aware of history. Lower
layer knowledge of wireless medium is shared with network
layer. Currently, path selection and spectrum selection are
common topics in cognitive routing protocols. Little prior
work has focused on multipath routing. We argue that
cognitive routing is a promising approach to make nodes
intelligently perform multipath routing.
III. 2BAPPROACH
In this section, we discuss the details of our proposed
routing protocol.
A. 8BOverview of CMMRP
The cognitive multipath multi-channel routing protocol
(CMMRP) is an on-demand disjoint multipath routing
protocol.
CMMRP lets nodes trigger routing updates reactively
when links break. Unlike the other on-demand multipath
routing protocols where multiple paths are discovered at once
by triggering one RREQ packet from source node, CMMRP
allows multiple paths to be discovered one by one by
triggering RREQ packets multiple times from the source node.
The benefits of modification are: