Opportunistic Routing in Multi-hop Wireless Networks *This material is mainly borrowed from authors and partially modified by the presenter Opportunistic Routing in Multi-hop Wireless Networks Sanjit Biswas and Robert Morris MIT CSAIL packet -> transmission (careful) Presented by Ying Huang CS598Hou 3/10/2006

ExOR: a New Approach to Routing in Multi-hop Wireless Networks mention last year we talked about low level link performance 1 kilometer Dense 802.11-based mesh Goal is high-throughput and capacity

Initial Approach: Traditional Routing packet packet A B src dst packet C just say picks a route, fwd over that route Identify a route, forward over links Abstract radio to look like a wired link

Radios aren’t Wires Every packet is broadcast src dst 1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4 5 5 5 5 6 6 6 6 C what’s really going on is we’re abstracting  as youall know, radios don’t work like links Every packet is broadcast Reception is probabilistic

ExOR: Exploiting Probabilistic Broadcast packet packet packet packet A B src dst packet packet packet packet packet C say there’s an opportunity here, take advantage of bcast. spell it out: closest node should reduce the number of tx to get to dest. challenge:: key failure is that two people should not forward. make sure only one guy forwards Decide who forwards after reception closest receiver to dst Challenge: agree efficiently and avoid duplicate transmissions

Outline Introduction Why ExOR might increase throughput ExOR protocol Measurements Related Work let’s assume we’re worked out the technical details (defer to design). let’s talk about why and how much of an improvement to expect.

Why ExOR Might Increase Throughput (1) Probabilistic Routing src N1 N2 N3 N4 N5 dst 75% 50% 25% new slide earlier, assume figured out design. explore why exor might improve throughput/develop a feel title -> why exor might beat trad routing. inverse prop to #trans, up to 4 hops Assumes probability falls off gradually with distance Best traditional route over 50% hops: 3(1/0.5) = 6 tx Throughput  1/# transmissions ExOR exploits lucky long receptions: 4 transmissions

Why ExOR Might Increase Throughput (2) Multi-path Diversity N1 25% 100% N2 25% 100% src dst 25% 100% N3 100% 25% N4 say this is a contrived example Traditional routing: 1/0.25 + 1 = 5 tx ExOR: 1/(1 – (1 – 0.25)4) + 1 = 2.5 transmissions Assumes independent losses

ExOR Uses Links in Parallel possible question – why are there only 7 forwarders.(just say we thin out...) Traditional Routing 3 forwarders 4 links ExOR 7 forwarders 18 links

Outline Introduction Why ExOR might increase throughput ExOR protocol Measurements Related Work

Node Agreement  Priority Ordering LIST: 4321 LIST: 4321 LIST: 4321 N2 N4 src dst N1 N3 LIST: 4321 LIST: 4321 LIST: 4321 Who is next hop? nodes “closest” to the destination send first To avoid collisions, one node sends at a time, ordered from higher priority to lower priority CLOSEST: Sort by ETX metric to dst Nodes periodically flood ETX “link state” measurements Path ETX is weighted shortest path (Dijkstra’s algorithm) Source sorts, includes a priority list of forwarder in ExOR header source calculates path etx for each node to dest, talk about dijkstra, sort by lowest etx

ExOR Batching Send batches of packets for efficiency rx: 53 rx: 85 rx: 99 rx: 40 rx: 22 rx: 23 rx: 57 rx: 88 rx: 0 N2 N4 tx: 0 tx: 100 tx:  9 tx: 57 -23  24 src dst N1 N3 tx:  8 tx: 23 Send batches of packets for efficiency Node closest to the dst sends first Other nodes listen, send remaining packets in turn Repeat schedule until dst has 90% of the whole batch Switch to traditional routing Cannot achieved amortized cost add circle for source, then some more circles showing next iteration (no rx/tx numbers). say at the end, packets go through multiple iteration. add source, add n4.

Reliable Summaries Batch Map Cumulative: what all previous nodes rx’d tx: {2, 4, 10 ... 97, 98} summary: {1,2,6, ... 97, 98, 99} Fragment N2 N4 src dst N1 N3 tx: {1, 6, 7 ... 91, 96, 99} Explore path diversity summary: {1, 6, 7 ... 91, 96, 99} Batch Map Repeat summaries in every data packet Cumulative: what all previous nodes rx’d Avoid duplicated retransmission This is a gossip mechanism for summaries just say we incude a summary in every packet repeated cumulative simplify: every node is listening, include list to make it more robust

Packet Scheduling Node transmission ordered from high priority to low priority Who initiates? Dst sends 10 empty packets with batch maps In absence of information , assume each higher priority node transmits at least 5 packets How to tell the end of fragment transmission Last packet in current fragment Timeout Estimate sender’s time to completion Sending rate of current sender & remaining packets

Scheduling Example

Protocol Design Layer 2.5 solution Batching: BatchSz BatchID PktNum Scheduling: FragNum FragSz Agreement: Batch Map

Using ExOR with TCP Web Server Client PC TCP ExOR Batches (not TCP) Node Gateway Proxy Web Proxy ExOR don’t say we will, say we do. Batching interacts badly with tcp. Internet cloud. NOT using TCP in the middle – spell it out Batching requires more packets than typical TCP window

Outline Introduction Why ExOR might increase throughput ExOR protocol Measurements Related Work

ExOR Evaluation Does ExOR increase throughput? When/why does it work well?

Evaluation Details 65 Node pairs 1.0MByte file transfer 1 kilometer 65 Node pairs 1.0MByte file transfer 1 Mbit/s 802.11 bit rate 1 KByte packets Traditional Routing ExOR 802.11 unicast with link-level retransmissions Hop-by-hop batching UDP, sending as MAC allows 802.11 broadcasts 100 packet batch size higher throughput implies less time spent on channel

ExOR: 2x overall improvement 1.0 0.8 0.6 Cumulative Fraction of Node Pairs 0.4 0.2 ExOR Traditional spend a little more time on the 240 x. 121. say this is just for the median, and it’s a factor of 2! 200 400 600 800 Throughput (Kbits/sec) Median throughputs: 240 Kbits/sec for ExOR, 121 Kbits/sec for Traditional

25 Highest throughput pairs 3 Traditional Hops 2.3x 2 Traditional Hops 1.7x 1 Traditional Hop 1.14x 1000 ExOR 800 Traditional Routing 600 Throughput (Kbits/sec) 400 for the two hop, say we can have more intermediate hops like ex 2 or make it all the way 200 Node Pair - Gossiped ACKs prevents incorrect retransmission - More available intermediate forwarders

25 Lowest throughput pairs 1000 ExOR 4 Traditional Hops 3.3x 800 Traditional Routing 600 Throughput (Kbits/sec) 400 200 etx was forced to use asymmetric links. robust batch maps. Node Pair Longer Routes - Utilize asymmetric link by gossiped batch map from different path

ExOR moves packets farther 58% of Traditional Routing transmissions 0.6 ExOR Traditional Routing Fraction of Transmissions 0.2 25% of ExOR transmissions 0.1 lower is better. right circle – using lots of longer links, sum them up and it’s 25%. so, like ex 1, using lots of long links. zeros: before many packets made no progress, with exor at least some. 100 200 300 400 500 600 700 800 900 1000 Distance (meters) ExOR average: 422 meters/transmission Traditional Routing average: 205 meters/tx - Utilize Long & Lossy Link

Future Work Choosing the best 802.11 bit-rate Cooperation between simultaneous flows Coding/combining

Summary ExOR achieves 2x throughput improvement ExOR implemented on Roofnet Exploits radio properties, instead of hiding them Discussion Memory requirement for packet buffering Jitters Instantaneous link quality N2 135 dst N1 Dst needs 3, but N2 has 3  Don’t send Need 3 ? 1346