1 Minimizing End-to-End Delay: A Novel Routing Metric for Multi- Radio Wireless Mesh Networks Hongkun Li, Yu Cheng, Chi Zhou Department of Electrical and Computer Engineering Illinois Institute of Technology Weihua Zhuang Department of Electrical and Computer Engineering University of Waterloo IEEE INFOCOM 2009

2 Outline  Introduction  End-to-end delay Metric  Achievable bandwidth along a multi-radio multi-channel path  Routing protocol design  Simulation  Conclusion

3  Routing in wireless mesh networks has been a hot research area in recent years  The multi-radio multi-channel connection has been widely considered as an efficient approach to increase the wireless network capacity  Most of the previous studies focus only on the transmission delay of the packet being served at the MAC layer Introduction

4 The transmission success probability The bandwidth of each link is 11Mbit/s The packet length is 1100bytes A transmission time is 0.8ms *10=9.6 ms 0.8*2+0.8* =11.2 ms Expected transmission time (ETT) End-to-End delay = 97.6 ms End-to-End delay = 24 ms Packet number in Queue

5 Goals  To develop a routing metric of end-to-end delay (EED) To monitor the transmission failure probability at the MAC layer to estimate the MAC transmission delay To count the number of packets waiting in the network layer buffer to estimate the queuing delay. The EED metric also implies the concept of load- balancing.

6 End-to-end delay Metric p i denote the transmission failure probability over link i L denote the packet length Bdenote link bandwidth W j denotes the contention window at the jth backoff stage the indicator I(A) is equal to 1 if A is true K is the maximum number of retransmissions.

7 End-to-end delay Metric

8 M i packets in the queue when a new packet reaches node n i

9 Achievable bandwidth over a multi-radio multi-channel path  Multi-Radio Achievable Bandwidth Inter-flow interference Intra-flow interference

10 Achievable bandwidth along a multi-radio multi-channel path Inter-flow interference the signal to interference and noise ratio (SINR) at receiver v >= the pre-determined threshold γ Notation: N denotes the background noise P v (u) the received power at node v from node u, v’ is the set of nodes located in interference range of v, P v (k) the interference power from an interfering node k

11 Achievable bandwidth along a multi-radio multi-channel path Inter-flow interference Interference degree ratio IDR i (uv) for link i between u and v P max is the maximum tolerable interference power at receiver v k1k1 k2k2 k3k3 k4k4 u IDR(uv)= / 5= 4/5 v k1k1 k2k2 u IDR(uv)=1+1 / 5= 2/5

12 Achievable bandwidth along a multi-radio multi-channel path Inter-flow interference Achievable bandwidth under the inter-flow interference (ABITF) Notation: B i denotes the physical bandwidth of link i ETX i denotes the expected transmission attempts to achieve a successful transmission over link i

13 Achievable bandwidth along a multi-radio multi-channel path Intra-flow interference Interference range within r (≥ 2) hops. Path 8 8 ABIRF(ij) denotes achievable bandwidth under the intra-flow interference (ABIRF) over links i and j 4 2 ABCDE channel Rate achievable bandwidth=4 mb/s 44

14 Achievable bandwidth along a multi-radio multi-channel path C D F EG source Destination Set ABITF=5=ABIRF AB Ch=1

15 Achievable bandwidth along a multi-radio multi-channel path C D F EG source Destination Set ABITF=5=ABIRF AB Ch=1 Ch=?

16 Achievable bandwidth along a multi-radio multi-channel path D F EG source Destination ABIRF=5 A Ch=1 Ch=2 ABIRF=3 C B

17 Achievable bandwidth along a multi-radio multi-channel path D F EG source Destination ABIRF=5 A Ch=1 ABIRF=3 ABIRF= (5*3 ) / (5+3) = 15/8 C B

18 Achievable bandwidth along a multi-radio multi-channel path source Ch=1 D F EG A C B Ch=2 Ch=3 3 Sub-paths ABIRF= 2 ABIRF= 1.8 ABIRF= 1.5

19 Achievable bandwidth along a multi-radio multi-channel path Weighted end-to-end delay Metric Notation: N P denotes the total number of packets queued in the buffers along the path L denote the packet length

20 Achievable bandwidth along a multi-radio multi-channel path Channel Diversity Coefficient Notation: B s denotes the achievable bandwidth of the same path if all links of this path work over the same channel, defined as the single-channel path capacity source Ch=1 D E A C B

21 Routing protocol design  Basic DSR Implementation  EED/WEED Based Routing

22 Routing protocol design_ DSR Source A K F Destination B C D E Addr A route request (RREQ) Addr A Addr B Addr A Addr B Addr C Addr A Addr B Addr C Addr D G route reply (RREP)

23 Routing protocol design_ EED Based Routing Source Destination Packet 1 Probe Message A has 2 Packet route reply (RREP) A B C DG Notation: λ = predetermined rate, M is the number of packets in its buffer carried in each probe V i denote the number of probes received from the upstream node associated with link i Packet 2

24 Routing protocol design_ WEED Based Routing Destination A B C DG channel ID IDR value A channel ID IDR value C Interface 1 Interface 2 EED

25 Simulation  A grid topology over a 1400m×1400m area  200m×200m square cells  Four flows are deployed at the 1st, 3rd, 5th, and 7th rows of the grid  Source/destination nodes of each flow located at both ends of the row  The other topology randomly places 40 nodes in a 1000m×1000m area  The queue size at each node is 20 packets  the link metric update interval in EED implementation is 50 seconds

26 Simulation_ EED-Based Routing in a Single-Channel Network

27 Simulation _ EED-Based Routing in a Single-Channel Network

28 Simulation _ EED-Based Routing in a Single-Channel Network

29 Simulation _ EED-Based Routing in a Single-Channel Network

30 Simulation_ WCETT Notation: β is a tunable parameter subject to 0 ≤ β ≤ 1. X j is the sum of transmission times of hops on channel j. The system has a total of k channels

31 Simulation the transmission failure probability interference range r = 1 hop channel assignment Packet Number

32 Simulation I > II IV > III I > II III > IV

33 Conclusions  The paper has key contributions in two aspects: The EED link metric considers both the queuing delay in network layer and transmission delay in the network layer A generic iterative approach is developed to compute the achievable bandwidth over a multi-radio multi-channel path  The designing link/path metrics that can lead to path selection with the minimum end-to-end delay and high network throughput

34 Thank you~

35 The forward delivery ratio, df, is the measured probability that a data packet successfully arrives at the recipient; the reverse delivery ratio, dr, is the probability that the ACK packet is successfully received.