1. Wireless Ad Hoc Network Routing Protocols
CSE
Maya Rodrig

2. Ad hoc networking
Infrastructureless networking – mobile nodes dynamically establish routing among themselves to form their own network on the fly.
Mobile nodes operate as routers
Mobile nodes participate in an ad hoc routing protocol
Mobile nodes operate as routers, forwarding packets for other mobile nodes that may not be within direct wireless transmission range.
Mobile nodes participate in an ad hoc routing protocol to discover ‘multi-hop’ paths through the network to any other node.

3. Why not reuse existing protocols?
Highly dynamic interconnection topology
LS generates loads of link status change msgs
DV suffers from out-of-date state or generates loads of triggered updates
Heavy computational burden on mobile nodes
Wireless medium differs in important ways from wired media
Existing protocols exhibit least desirable behavior when presented with a highly dynamic interconnection topology.
Wireless media are of limited and variable range.

4. The Protocols DSDV, TORA, DSR, AODV Proactive vs. reactive (on-demand)
Destination based: node maintains only local topology info, no global view
Proactive: table driven; continuously evaluates routes; no latency in route discovery; periodic control messages needed; large network capacity to keep information up-to-date; most routing info may never be used
Reactive: on demand; route discovery by global search; drawback-latency of route discovery; may not be appropriate for real time communication

5. Destination-Sequenced Distance Vector (DSDV)
Preserve the simplicity of RIP while avoiding the routing loop problem
Hop-by-hop distance vector
Routing table contains entries for every reachable node
Each route is tagged with a sequence number originated by destination (even numbers)
Routing info is transmitted by broadcast
Updates are transmitted periodically and when there is a significant topology change
DV – Distributed Bellman-Ford (DBF), Routing Information Protocol (RIP)
Hop-by-hop DV
Each route is tagged with a sequence number originated by destination so nodes can quickly distinguish stale routes from new ones and avoid formation of routing loops.

6. DSDV cont.
Route R is more favorable than R’ if R has a greater sequence number or if the two routes have equal sequence numbers but R has a lower metric (hop count)
Broken links are indicated by “” metric and the sequence number of destination is incremented to odd number before broadcast

7. No count to infinity

8. Temporally-Ordered Routing Algorithm (TORA)
Based on a “link-reversal” algorithm
Node broadcasts a QUERY packet which propagates to destination or to node having a route to the destination
Recipient of the QUERY broadcasts an UPDATE packet listing its height with respect to the destination
Each node that receives the UPDATE sets its height to be greater than the height of the neighbor from which the UPDATE came  creates a series of directed links from the QUERY originator to the node initiating the UPDATE
Based on a link reversal algorithm; on demand; provides multiple routes; minimize communication overhead by localizing algorithmic reaction to topological changes when possible
Route optimality (shortest path) is considered of secondary importance

9. TORA cont.
When a node discovers a route is no longer valid, it adjusts its height so that it is a local maximum and transmits an UPDATE
When a network partition is detected, a node generates a CLEAR packet to reset routing state and remove invalid routes

10. Dynamic Source Routing (DSR)
Packet headers contain the route the packet must follow
Route Discovery:
Source node S broadcasts Route Request packet that is forwarded through the network
Destination node D or another node that knows a route to D answers with a Route Reply
Route Maintenance:
When the network topology has changed s.t. the route to D can no longer be used, a Route Error packet is sent to S
S can try another route to D from its cache or invoke Route Discovery again
Network interfaces in promiscuous mode  nodes cache overheard route information
intermediate nodes don’t need to maintain up-to-date info

11. DSR Example

12. Ad Hoc On-Demand Distance Vector (AODV)
Combination of DSR (on demand) and DSDV (hop-by-hop routing, sequ nums)
Node S broadcasts a Route Request message for destination D, including the last known sequence number for D
Node with a route to D generates a Route Reply with its sequence number for D
Nodes that forward Route Request store reverse route back to S; nodes that forward Route Reply store forward route to D

13. AODV cont. No HELLO messages from neighbor indicate link is down
Nodes that recently forwarded packets using the failed link are notified via an UNSOLICITED ROUTE REPLY with infinite metric for the destination  reinitiate Route Discovery

14. Simulation Environment
Model attenuation of radio waves between antennas
Link layer implements standard MAC protocol DCF
Broadcast packets sent only when virtual and physical carrier sense indicate the medium is clear (no RTS/CTS and no ACKs)
DCF = Distributed Coordination Function to accurately model the contention of nodes for the wireless medium

15. Methodology Network simulation Movement model Communication model
50 wireless nodes moving in 1500m*300m flat space
Over 200 different scenarios
Movement model
“Random waypoint” model (pause times: 0, 30, 60, 120, 300, 600, 900 seconds)
Avg speed 10 meters/second
Communication model
Sending rates: 1, 4, 8 packets/second
10, 20, 30 CBR sources
Packet size of 64 bytes
Goal: measure the ability of the routing protocols to react to network topology change while continuing to successfully deliver data packets to their destinations
Scenarios = workloads, conditions
Simulations last 900 seconds. Pause time 0 = continuous motion = no motion
CBR = constant bit rate

16. Metrics
Packet delivery ratio- ratio between num packets originated by sources and num packets received at their destination
Routing overhead- num routing packets transmitted during the simulation
Path optimality- difference between the num hops a packet took to reach its destination and the length of the shortest path
Overhead – each hop in a multi-hop transmission counts as one transmission
Optimality – length of shortest path that physically existed in the network when the packet was originated

17. Packet Delivery Ratio DSR and AODV deliver over 95% of data packet
TORA does well with 20 sources
DSDV fails to converge at pause time < 300
20 sources
DSDV: packets are lost because stale routing table entry directed them to be forwarded over a broken link. Maintains only one route, so undelivered packet is dropped since there are no alternate routes.
TORA: packet drops dues to the creation of short lived routing loops that are a natural part of its link-reversal process

18. Routing Overhead TORA, DSR, AODV are on demand
DSDV is largely periodic
DSR limits overhead of Route Requests through caching
20 sources
On demand: overhead drops as mobility drops. as num sources increases -> num routing packets increases because more there are more destinations to which the network must maintain working routes
DSVD: overhead is nearly constant with respect to mobility rate. Each destination broadcasts a periodic update with a new sequence number every 15 seconds. With 50 unsynch nodes, at least one node broadcasts a periodic update during each second
TORA requires about 8 times the overhead of DSR. However if routing overhead is measured in bytes, including bytes of the source route header that DSR places in each packet, DSR becomes more expensive that AODV

19. Path Optimality
Internal mechanism knows the length of the shortest path between all nodes at any time
DSDV and DSR use routes close to optimal
AODV and TORA have a tail

20. Another Protocol: Greedy Perimeter Stateless Routing (GPSR)
Geography to achieve scalability in wireless routing protocols
Assume bidirectional radio reachability
Assume a location registration and lookup service that maps node addresses to locations
Position of a packet’s destination and positions of candidate next hops sufficient to make correct decisions
Scalability under increasing number of nodes and increasing mobility rate

21. Greedy Forwarding
Beaconing algorithm provides all nodes with their neighbor’s positions
Packets are marked with their destinations’ locations
A forwarding node makes a locally optimal greedy choice: next hop is the neighbor geographically closest to the destination
Problem: topologies in which the only route to the destination requires temporarily moving farther in geometric distance from the destination

22. Planar Perimeters
Right-hand rule : when arriving at node x from node y, the next edge traversed is the next one sequentially counterclock-wise about x from edge (x,y)  navigating around the void
Construct planarized graphs to eliminate crossing links from the network without partitioning the network

23. GPSR versus DSR
Packet Delivery Success Rate
Routing Overhead
50 nodes

24. Comparison cont.
Path Length
Network Diameter

25. Choosing Routes
Shortest path is not a good metric  choose routes with less capacity than best existing paths
Minimum hop-count routes include links with high loss ratios  retransmissions consume bandwidth
MIT
HotNets I 2002

26. Link Behavior in Experimental Networks
Link quality distribution is spread out
30% of link pairs are unusable
Best 40% of link pairs deliver 90% of their packets
30% link pairs have asymmetric delivery rate
Delivery rates sometimes change very quickly (averaging not applicable)
No good correlation between delivery rate and radio’s signal strength
We need practical estimates for link quality and ways to combine link metrics into path metrics
Static wireless networks:
Indoor network: 18 nodes
Rooftop network: 5 nodes

27. Expected Transmission Count (ETX)
Find paths with fewest expected number of transmissions required to deliver a packet to its destination
Use per-link measurements of delivery ratios in both directions
Modified DSDV and DSR
ETX outperforms minimum hop-count
ETX incurs more overhead due to loss-ratio probes
MobiCom 2003
Delivery ratios are measured using dedicated link probe packets
Forward and reverse delivery ratio
Identify paths with high end-to-end throughput

28. …
Early protocols assume cooperating nodes that are willing to forward packets for others
The role of power in routing protocols