Directional neighbor discovery in dual-band systems

The neighbor discovery process is a key component of Device-to-Device (D2D) communication. This paper considers the problem of directional neighbor discovery in dual-band systems, where wireless devices operate in both mm-wave and microwave frequency bands and employ antenna arrays. Due to the propagation loss, the set of neighbors in a direction at mm-wave frequencies is a subset of the neighbors at microwaves. Consequently, the knowledge of neighbors in one of the bands allows to greatly accelerate the neighbor discovery process. We propose a distributed random discovery algorithm for the mm-wave band, where each device finds the relevant algorithm parameters, i.e., transmission and beam steering probabilities, given only local information at the microwave band. Because each node knows only its own local information, imposition of a fairness condition prevents selfish behavior that would deteriorate network performance. Optimum algorithm parameters are obtained as solutions of a linear program. Simulations show that the proposed scheme outperforms the optimum single-band random scheme and significantly reduces the average discovery time, especially for small numbers of neighbors per beam.