Impact of Interference on the Throughput of a Multihop Path in a Wireless Network

1. Impact of Interference on the Throughput of a Multihop Path in a Wireless Network Amit K. Vyas (amitvyas@cs.stanford.edu) Fouad A. Tobagi (tobagi@stanford.edu) Presented at BROADNETS 2006 (Recipient of the Best Wireless Paper Award) Wireless Mesh Networks (AIIT)

2. 802.11 WIRELESS NETWORK Wireless Stations or Relays Motivation Issues: • Multiple wireless nodes interfere with each other, resulting in: • CSMA Blocking of Nodes  Reduces no. of chances to attempt transmission • Simultaneous Transmissions by Many Nodes  Lower SINR  Higher PER • How does this affect the throughput of an end-to-end path ? • What is the impact of interference due to packets of the same end-to-end flow (intra-path) ? and due to packets of different flows (inter-path) ? • What is the sensitivity to choice of transmission power and MAC parameters (Energy Detect threshold) ? Wireless Mesh Networks (AIIT)

3. d Scenario • Scenario for Intra-Path Interference: Pt, ED, R Pt = Transmission Power (dBm) ED = Energy Detect Threshold (dBm) R = Data Rate (Mbps) d = Distance between nodes (meters) • For illustration, we consider a single multihop path of uniformly spaced nodes in an ad hoc wireless network • Saturated Traffic flowing from one end to another • Same Propagation Characteristics Wireless Mesh Networks (AIIT)

4. Carrier Synchronization Physical Layer (PLCP) Header Data Reception (CRC check) System Model • Receiver: • Models three stages of packet reception • Fading:Average PER vs. SNR curves generated by simulation of various stages of the 802.11a transmitter and receiver [2][3] – used to model PLCP and data errors, while synchronization is modeled using a fixed threshold • Above models incorporated into simulator based on GloMoSim Wireless Mesh Networks (AIIT)

5. 0 6 7 8 1 2 9 5 S0 1362 S1 944 S5 525 S6 497 S7 497 S8 497 S2 809 > > > > = =      0 0 0 Illustration Throughput Bottleneck Si = Throughput from node i to i+1 in kbps γ = 4.1 d = 10 m Pt = 29.03 dBm R = 6 Mbps ED = -91 dBm Sync = -85 dBm • Link 6-7 became the bottleneck because of interference from both sides • Nodes 7 and beyond have almost empty queues, but cause sufficient interference to result in a bottleneck at link 6-7 Wireless Mesh Networks (AIIT)

6. Effect of Transmission Power γ = 4.1 d = 5 m ED = -91 dBm Sync = -85 dBm Poor Link Performance Excessive Blocking • High SINR, but low number of simultaneous transmissions Wireless Mesh Networks (AIIT)

7. Effect of Energy Detect Threshold γ = 4.1 d = 5 m Pt = 20 dBm Sync = -85 dBm Excessive Blocking • Low number of simultaneous transmissions Excessive interference • due to high number of simultaneous transmissions Wireless Mesh Networks (AIIT)

8. Effect of Path Loss Exponent • Intra-path Interference – limiting factor in low-attenuation environments • Transmission Power – limiting factor in high-attenuation environments • Throughput is optimized over Pt , R, ED • Pt < 29 dBm as per 802.11a standard Wireless Mesh Networks (AIIT)

9. Where is the Throughput Lost ? • Max throughput over all γ is around 4.5 Mbps – less than 10% of the maximum PHY data rate of 54 Mbps !!! • To understand where it is lost, consider a perfectly scheduled TDMA scheme with optimum transmission power ( ) • Even then the maximum throughput is only 9.2 Mbps !!! Wireless Mesh Networks (AIIT)

10. Where is the Throughput Lost ? • γ = 4.1 • Sync = -85 dBm • Optimum values of Pt, R and MAC parameters (ED Threshold and TDMA Interval) • Intra-path Interference – Reason for most of the loss in multi-hop throughput • Transmission Power - Limits throughput in high attenuation environments Wireless Mesh Networks (AIIT)

11. Comparison of MAC Schemes • γ = 4.1 • Optimum values of Pt , R and MAC parameters • IEEE 802.11 throughput 40% lower than TDMA • Optimum values of 802.11 and TDMA parameters related – The number of nodes blocked on an average using 802.11 corresponds to the TDMA interval Wireless Mesh Networks (AIIT)

12. Impact of Inter-Path Interference on Throughput • Higher separation between the paths  decrease in inter-path interference  more simultaneous transmissions and higher throughput • No Additional loss of throughput due to inter-path interference – only sharing of throughput between multiple paths Wireless Mesh Networks (AIIT)

13. Conclusions • Intra-path Interference – Most important factor limiting throughput of multihop paths to less than 10% of that over a single link. • Low-Attenuation Environments(γ ≤ 2) are more affected because signals decay very slowly with distance • Inter-path Interference – Results in sharing of throughput between multiple paths, but no further loss of throughput • MAC Efficiency– IEEE 802.11 throughput around 40% lower than that of an optimally scheduled TDMA scheme • Sensitivity to Parameters– Throughput can be substantially lower if Pt, R and ED are not tuned properly Wireless Mesh Networks (AIIT)

14. References • J. Medbo and P. Schramm, “Channel models for hiperlans/2 in different indoor scenarios,” HIPERLANS/2 ETSI/BRAN contribution, Mar. 1998. • O. Awoniyi and F. A. Tobagi, “Packet error rate in OFDM-based wireless LANs and QoS for VoIP and TCP based applications,” in Proc. INFOCOM 2006. Twnety-Fifth Annual Joint Conference of the IEEE Computer and Communications Societies, 2006. • O. Awoniyi and F. A. Tobagi, “Effect of fading on the performance ov VoIP in IEEE 802.11a WLANS,” in Proceedings of the IEEE International Conference on Communications, vol. 6, June 20-24, 2004, pp. 3712-3717. This talk published as: A. K. Vyas and F. A. Tobagi, “Impact of Interference on the Throughput of a Multihop Path in a Wireless Network,” in Proc. IEEE BROADNETS 2006. Third International Conference on Broadband Communications, Networks and Systems, San Jose, California (USA), Oct 1 – 5, 2006. (Recipient of the Best Wireless Paper Award) Wireless Mesh Networks (AIIT)

15. Thank You !! Wireless Mesh Networks (AIIT)