Neighbor discovery in wireless networks: A multiuser-detection approach

The problem of determining which sensors are neighbors of a given one in a wireless network operating on a frequency-flat Gaussian channel is considered. While recent neighbor discovery (ND) algorithms operate at protocol level, the developed schemes exploit signal processing techniques to perform ND at physical layer. Similarly to the existing approaches, it is assumed that sensors transmit at random instants copy of the signature waveform associated to their identities. Borrowing techniques from multiuser detection, it turns out that the developed ND algorithms can handle multi-sensor interference environments where protocol level algorithms entail waste of energy due to packet collisions. A number of neighbor discovery algorithms based on different optimization criteria are introduced and analyzed. Numerical results show that reduced complexity algorithms can achieve a satisfactory performance.