Oblivious Rendezvous in Cognitive Radio Networks

Abstract

Rendezvous is a fundamental process in the operation of a Cognitive Radio Network (CRN), through which a secondary user can establish a link to communicate with its neighbors on the same frequency band (channel). The licensed spectrum is divided into N non-overlapping channels, and most previous works assume all users have the same label for the same channel. This implies some degree of centralized coordination which might be impractical in distributed systems such as a CRN. Thus we propose Oblivious Rendezvous where the users may have different labels for the same frequency band.

In this paper, we study the oblivious rendezvous problem for M users (ORP-M for short) in a multihop network with diameter D. We first focus on the rendezvous process between two users (ORP-2) and then extend the derived algorithms to ORP-M. Specifically, we give an Ω(N ²) lower bound for ORP-2, and propose two deterministic distributed algorithms solving ORP-2. The first one is the ID Hopping (IDH) algorithm which generates a fixed length sequence and guarantees rendezvous in O(N max {N,M}) time slots; it meets the lower bound when M = O(N). The second one is the Multi-Step Hopping (MSH) algorithm which guarantees rendezvous in O(N ² log _N M) time slots by combing ID scaling and hopping with different steps; it meets the lower bound if M can be bounded by a polynomial function of N, which is true of large scale networks. The two algorithms are also applicable to non-oblivious rendezvous and the performance is comparable to the state-of-the-art results. Then we extend the algorithms to ORP-M with bounded rendezvous time by increasing the diameter D by a factor.