november 4, 2003applied research laboratory, washington university in st. louis apoc 2003 wuhan,...
DESCRIPTION
APOC 2003 Wuhan, China Applied Research Laboratory, Washington University in St. Louis 3/10 Introduction Wireless networks become increasingly popular and affordable Ad hoc mobile wireless networks –Dynamic and easy deployment –No fixed infrastructure required –Applications Emergency disaster relief, battlefields, information gathering, group interaction Routing is crucial to the success of ad hoc mobile wireless networksTRANSCRIPT
November 4, 2003 Applied Research Laboratory, Washington University in St. Louis
APOC 2003Wuhan, China
Cost Efficient Routing in Ad Hoc Mobile Wireless Networks
Ruibiao [email protected]
Applied Research Laboratory, Washington University in St. Louis
2/10APOC 2003Wuhan, China
Outline
• Background and motivation• Cost model in ad hoc networks • Cost-based routing scheme• Adaptation to existing routing protocols• Experimental results• Conclusions
Applied Research Laboratory, Washington University in St. Louis
3/10APOC 2003Wuhan, China
Introduction• Wireless networks become increasingly popular
and affordable• Ad hoc mobile wireless networks
– Dynamic and easy deployment– No fixed infrastructure required– Applications
• Emergency disaster relief, battlefields, information gathering, group interaction
• Routing is crucial to the success of ad hoc mobile wireless networks
Applied Research Laboratory, Washington University in St. Louis
4/10APOC 2003Wuhan, China
Routing in Ad Hoc Networks• Select a suitable path between two nodes • Routing challenges in ad hoc networks
– Dynamics, limited resources, environment complexity• Table-driven proactive protocols
– Maintain consistent up-to-date route information of the network with constant updates
– Examples: DSDV, CGSR, WRP• Source-initiated on-demand reactive protocols
– Invoked only when there is a data delivery request but no existing paths– Examples: AODV, DSR, TORA, ABR
• Location-based protocols– Use information about destination location to forward data packets– Examples: LAR, GPSR, SPEED
Applied Research Laboratory, Washington University in St. Louis
5/10APOC 2003Wuhan, China
Ad Hoc Networks Cost Model • Traditional protocols use hop count only as routing
metrics– Unbalanced link bandwidth– Hops distance– Data forwarding is more costly than wired networks
• Observation: traffic variation diminishes as flows aggregate
• More precise cost-based model is desirable– Consolidate all factors– Take advantages of traffic aggregation
• Higher degree of traffic variation in wireless networks
Applied Research Laboratory, Washington University in St. Louis
6/10APOC 2003Wuhan, China
Cost Model• Connection cost depends on the individual flows• Individual flows exhibit statistically similar behavior• Aggregated flow variation is smaller than the sum of all
individual flow variationCu,v(f) = lu,v ( f + k f ½)Cp(f) = {(u,v) p}Cu,v(f) = {(u,v) p} lu,v ( f + k f ½)– Cu,v(f): cost of transmitting flow volume f between nodes u and v– Cp(f): cost of transmitting flow volume f on path p– lu,v: length between nodes u and v– f: total traffic flow load : individual traffic variation parameter– k: constant factor
Applied Research Laboratory, Washington University in St. Louis
7/10APOC 2003Wuhan, China
Cost-based Routing Scheme• Use the ad hoc network cost model to replace
the hop count as the routing metrics• Basic operations
– Cost computation• By the cost model
– Cost update• Per-link incremental cost Cu,v incurred by flow increment
Cu,v = lu,v [ + k (f + )½ - k ½)]
• Whole path incremental cost Cu,v is computed similarlyCp = {(u,v) p} Cu,v = {(u,v) p} lu,v [ + k (f + )½ - k ½)]
Applied Research Laboratory, Washington University in St. Louis
8/10APOC 2003Wuhan, China
Adaptation to Existing Routing Protocols• Proactive Routing Protocols
– Use new routing tables with Cu,v(link state protocols) or Cp(distance vector protocol) on nodes
• Reactive Routing Protocols– Route discovery
• Use Cu,v to determine the optimal among all outgoing links • Forward requests with piggybacked increment cost
information along all paths, and the destination makes the final decision
– Route maintenance• Changes trigger updates to neighbors
• Location-based Routing Protocols– No need to record distance information
Applied Research Laboratory, Washington University in St. Louis
9/10APOC 2003Wuhan, China
Experimental Results• ns-2 simulator with wireless extension
– Varied number of nodes (50 –200)– IEEE 802.11 radio– Modified vs original AODV
• Total cost comparison
AODV-cmod
AODV-orig
Applied Research Laboratory, Washington University in St. Louis
10/10APOC 2003Wuhan, China
Conclusions
• Ad hoc mobile wireless networks present great challenges to traditional routing algorithms
• A cost model for ad hoc mobile wireless network proposed
• Cost-based routing scheme is described – Cost efficient routing– Adaptive to existing routing protocols