Copyright Intel Corporation 2000 -2004 Making Radios More

Copyright© Intel Corporation 2000 -2004 Making Radios More Like Human Ears Lakshman Krishnamurthy Principal Engineer Intel Corp. Jing Zhu, Xingang Guo, L. Lily Yang, W. Steven Conner Intel Corp. ®

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Three points We can improve performance by making radios like our ears And behaving like people – talk even though you hear others Mesh can give more bandwidth – not less Stop working on routing and NS 2! 2 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Problem overview Mesh network An ad-hoc group of nodes relaying each other’s traffic Logically flat hierarchy – AP mesh, station mesh, hybrid mesh Spatial reuse – use the same channel at spatially separated locations Enable simultaneous communications to improve overall network throughput Applicable to large-scale wireless networks * Third party brands/names are property of their respective owners 3 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Physical carrier sensing 802. 11 MAC based on CSMA/CA CS (Carrier Sensing) to avoid interference Carrier sensing – a station listens before transmit Listen – sample the radio energy (interference) in the air Carrier sensing threshold Decide transmit or wait Current devices – static, not independently tunable Make threshold tunable, and network throughput can be improved dramatically with properly tuned threshold 4 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Network throughput Large-scale 802. 11 networks, in each channel Link date rate – R 11 Mbps # of simultaneous comm. – N 10 X) . Network throughput (R*N) 110 Mbps “N” determined by spatial reuse Reuse the same channel in separated location Tuning CS threshold can leverage spatial reuse 5 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Communication model Each data rate has its own requirement on channel quality SNIR threshold Spatial reuse Properly separate simultaneous comm. Different rates will require different separation distances CS threshold reflects separation distance 6 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Anatomy of interference B X C D R TX I RX 7 A Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Simulating chain network 90 -node chain, 90 hop e 2 e path Tx range tuned to node distance Measure e 2 e throughput while varying Pcs_t E 2 e throughput changes dramatically Optimal Pcs_t depends on data rate 9 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Simulating grid network 10 x 10 grid, comm. w/ immediate neighbors Tx range tuned to node distance Measure aggregate throughput while varying Pcs_t E 2 e throughput changes dramatically Optimal Pcs_t NOT depending on propagation environment 11 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 B RTS/CTS? C Protocol exchange may fail when outside of Tx range D R TX RX I A VCS failed to take full advantage of higher pathloss to increase spatial reuse 12 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Conclusion Properly tuned carrier sensing achieves optimal spatial reuse Dramatically improves network throughput Computational efficient Complementary to RTS/CTS Non-disruptive enhancement to 802. 11 MAC Make the carrier sensing tunable in all 802. 11 devices 13 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Overview: Experimental evaluation of an 802. 11 b home mesh network 71 70 B 73 A Office Den 72 74 C Back Yard D 77 Living Room 75 76 Lower Level Upper Level Experiments performed in house (~2000 sq. ft. ) in Hillsboro, OR (August, 2003) Topology: 8 Client Laptops and 4 AP routers In a real home network scenario, some of the laptops would likely be replaced by other 802. 11 enabled devices (e. g. , DVRs, media servers, stereo systems, etc. ) Traffic: Experiments assume network traffic is not limited to Internet surfing on a broadband link Clients share significant amount of data within the home (e. g. , A/V content sharing, photo storage, data backup, etc. ) 14 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Individual Node Throughput Non-Mesh BSS Individual Node Throughput Multi-Hop ESS Individual Node Throughput 5 Throughput (Mbps) 6 4 70 (O) 3 73 (D) 5. 182 75 (L) 2 77 (B) 1 1. 7 X 3. 1 X 4 3 Connected! 70 (O) 5. 179 73 (D) 75 (L) 2 77 (B) 2. 679 2. 686 1. 8 1 1. 572 0. 85 0 0 Office t Ou ra of ng Living Room Den 0 Backyard Office Living Room Den Backyard e 15 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Multi-Node Throughput Non-Mesh BSS Aggregate Throughput Multi-Hop ESS Aggregate Throughput 5. 338 5. 322 1. 3 X 3. 910 2. 878 1. 994 t Ou ra of ng 1. 9 X 2. 1 X 3. 880 3. 284 1. 520 e 16 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Multi-Node Throughput cont. Aggregate Throughput with 8 Clients 2. 1 X 3. 709 1. 719 t Ou ra of ng e 17 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Client-to-Client Throughput Non-Mesh BSS Client-to-Client Throughput Multi-Hop ESS Client-to-Client Throughput 3. 4 X 2. 4 X f to ran ge Ou • Note: Direct client-to-client links can help here as well 18 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Network Latency Non-Mesh BSS End-to-End Latency Multi-Hop ESS End-to-End Latency ~ 2 ms increase per hop n f ra ge to Ou • Highly dependent on implementation 19 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Shorter range radio hops offer higher throughput Copyright© Intel Corporation 2000 -2004 Source: Intel Corporation 20 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Summary of home testbed Results Copyright© Intel Corporation 2000 -2004 A multi-hop mesh is beneficial, even for a relatively small-scale home network Multi-hop topologies: Can be built with standard 802. 11 hardware Can improve network performance in comparison to traditional 1 -hop BSS networks These experiments used 1 radio on each AP/router; multi-radio per AP/router would allow even better performance (multi-channel) 21 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Mesh test bed and Platforms • 25 -35 nodes • Laptops and embbeded Xscale boards (PXA-255 and IXP 425) • Boards, software available for research • Performance comparison of mesh and wireless network self organization 22 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Topology and Transmit power have significant impact Microsoft Research Routing Protocol LQSR w/ shortest path (No link quality metric) LQSR w/ ETX (MIT 2002) LQSR w/ ETX++ (MSR 2003) AODV w/ link filtering (Intel 2003) Median Throughput of 100 TCP transfers Intel Routing Protocol Median Throughput of 100 TCP transfers LQSR w/ shortest path 1155 Kbps (No link quality metric) LQSR w/ ETX 1379 Kbps 7334 Kbps (MIT 2002) 1935 Kbps LQSR w/ ETX++ 1601 Kbps (MSR 2003) AODV w/ link filtering 1460 Kbps 11709 Kbps (Intel 2003) 2079 Kbps Need self-configuration algorithms Communications Technology 23 Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 802. 11 S MESH standard Lowering the Barriers to 802. 11 Mesh Deployment Parallel Efforts: Standardize Multi-Hop ESS Mesh Interoperability Radio/Metric-Aware L 2 Routing/Switching Security Self-Configuration / Management Enhance MAC Performance for Mesh Scalability Scheduling (managing collisions/ interference) Major focus of new Mesh Task Group (802. 11 s) Leverage 802. 11 i/k where possible Influence current/ future MAC enhancement efforts to improve scalability for mesh Leverage 802. 11 e/n where possible Mesh-specific MAC enhancements can be made in ESS Mesh TG 24 Communications Technology Lab

Leveraging spatial reuse with enhanced physical carrier sensing Copyright© Intel Corporation 2000 -2004 Thank you 25 Communications Technology Lab