Have You Recorded My Voice: Toward Robust Neighbor Discovery in Mobile Wireless Networks

Abstract

The surge of proximity-based applications on mobile devices has promoted the need for effective neighbor discovery protocols in mobile wireless networks. In contrast to existing works, which can achieve energy efficient neighbor discovery with bounded latency only in the scenario without strong interference, we aim at designing techniques for practical and robust neighbor discovery. We propose ReCorder to achieve robust neighbor discovery in mobile wireless networks despite the “noisy” communication media. Specifically, we exploit the cross-correlation property of pseudo-random sequences to eliminate the necessity of beacon decoding in existing neighbor discovery protocols. In ReCorder, a neighbor discovery message can be detected through cross-correlation on an RCover preamble, and contains a ReCord identity signature, which is unique for each of the nodes. We also design algorithms for RCover detection and ReCord recognization. The performance of the ReCorder has been evaluated using the USRP-N210 testbed. Our evaluation results show that the ReCorder can achieve robust neighbor discovery at an SINR lower than the existing beaconing and decoding-based neighbor discovery protocols by almost 10 dB. Furthermore, the ReCorder can avoid degrading the decoding of background IEEE 802.11 a/g transmissions with BPSK modulation, which is important for its co-existence with concurrent wireless streams, and it only induces limited throughput degradation to background data flows.