Neighbor discovery in peer-to-peer wireless networks with multi-channel MPR capability

We study the time duration of neighbor discovery in peer-to-peer wireless networks with multi-channel multi-packet reception capability. Radios with this capability, at any time slot, can either transmit messages in one channel or receive messages from all channels. This capability is provided by emerging CDMA/OFDMA techniques. Neighbor discovery in this scenario is different from other multi-packet reception scenarios like MIMO since collision can still happen in single channel. To discover neighbors with such capability in single-hop networks, we prove that the expected running time of all randomized algorithms is Θ(n/k) where n is the number of nodes and k is the number of channels. We provide a Las Vegas algorithm that finds all neighbors with variable running time. We prove that its running time is Θ(n/k ln n) with high probability. We also give a Monte Carlo algorithm that terminates in Θ(n/k) time. We show that it finds all neighbors with high probability. Both algorithms are validated by simulation.