FastND: Accelerating Directional Neighbor Discovery for 60-GHz Millimeter-Wave Wireless Networks

摘要

Neighbor discovery (ND) is a critical primitive for 60-GHz wireless networks with highly directional radios. Prior work has attempted to improve the ND efficiency but overlooks the unique properties of 60-GHz phased-array antennas and spatial channel profile. In this paper, we first conduct a systematic study of the ND problem using a reconfigurable 60-GHz radio. Combined with an analytical model, we find that environmental characteristics and client mobility substantially affect 60-GHz ND latency, and due to inherent spatial channel sparsity of 60-GHz channels, even short-distance links can experience intolerable latency. To solve these new challenges, we propose a mechanism called FastND that accelerates ND by actively learning the spatial channel profile. FastND leverages steerability of 60-GHz phased-array antennas and accumulates channel information by overhearing beacon preambles along different beam directions. Using a compressive sensing framework, together with a strategical beam selection mechanism, FastND can infer the strongest spatial angle to listen to, thereby increasing the likelihood to quickly decode beacons and achieve ND. Our testbed experiments and ray-tracing tests demonstrate that FastND can reduce 802.11ad ND latency to 1/10-1/2, with different levels of mobility, human blockage, environmental sparsity, and non-line-of-sight links.