Practical Multi-antenna Spatial Reuse in WLANs

Sriram Lakshmanan1, Karthikeyan Sundaresan2, Mohammad Khojastepour2, and Sampath Rangarajan2
1GNAN Research Group
School of Electrical and Computer Engineering
Georgia Institute of Technology
2NEC Labs. America
Princeton, NJ

The 7th International ICST Conference on Broadband Communications, Networks, and Systems (BROADNETS 2010), Athens, Greece, October 25-27, 2010.


Smart antennas can improve spatial reuse in a wireless network through interference suppression. However, interference suppression requires support from clients in the form of channel estimation, which existing clients do not support. In this work, we explore practical solutions to obtain spatial reuse with smart antennas without requiring hardware changes to clients. To this end, we design a novel solution for improving spatial reuse in indoor WLANs which uses ‘approximate’ channel estimates and still yields close to ideal performance. Our solution called Light-weight Multi-Antenna Spatial Reuse (LSR) consists of (i) a multilink channel estimation scheme that can be realized with simple Received Signal Strength (RSSI) measurements that existing WLAN clients provide readily, (ii) a low-complexity scheduler to decide the subset of beamformed links that must be active concurrently. We demonstrate that the estimates obtained using this scheme when used with a multi-link beamforming technique such as Zero Forcing yields significant interference suppression benefits. We implement the channel estimation scheme on a testbed of software radio clients to demonstrate its practical feasibility. Further, we evaluate the performance of LSR using extensive signal strength traces from 802.11g Access Points equipped with eight element antenna arrays in an indoor office environment. The results indicate that LSR achieves close to the performance obtained with an optimal scheme that uses accurate channel estimates and also improves the median sum rate of indoor users by up to 2.7x over competing approaches.

Presentation: []      Full Paper: [pdf (289KB)]