Resource Allocation for Cognitive Satellite Communications Downlink

Cognitive satellite communications (SatCom) is considered to be able to alleviate the bottleneck of spectrum resource shortage due to traditional spectrum allocation. This paper focuses on a special scenario where the frequency band of SatCom is recommended by the terrestrial terminal according to its spectrum sensing results. Further speaking, frequency bands preferred by terminals in each coverage beam of the satellite may be random, which on the whole forms diverse recommended channels problem that poses a great challenge to traditional multi-beam satellites. To make reasonable use of available resources in this scenario, this paper targetedly proposes a beam hopping (BH) scheme, which is capable of providing services on each frequency band. Based on the BH scheme, two 4-D [i.e., time, frequency, power, and dedicated spot (DS) beam] resource allocation (RA) schemes are presented, which adopt maximizing throughput (MT) and minimizing demand-supply variance (MDSV) as objectives, respectively, corresponding to the fact that satellite resources may be relatively rich or scarce. Both of the RA problems belong to mixed-integer nonlinear programming. By decomposing them, three levels of problems, namely, frequency band selection (FBS) problem, dedicated beam allocation (DBA) problem, and time-power allocation (TPA) problem are successively formed. For the FBS and DBA problems, we correspondingly propose heuristic algorithms to quickly distribute frequency bands and dedicated beams. Whereas for the TPA problem, Lagrangian dual algorithm and water-filling-assisted Lagrangian dual algorithm are respectively adopted to solve the convex problem for MT and the nonconvex problem for MDSV. Taking also spectrum sensing errors into account, numerical simulations show that the proposed schemes and algorithms perform well, and a significant gain can be achieved in cognitive SatCom.