Resource Allocation in OFDMA Cognitive Radio Systems Based on Ant Colony Optimization

Traditional spectrum management, in order to guarantee the coexistence of the various wireless services, assigns prescribed and fixed frequency bands to licensed users, or, primary users (PUs). Anyway these portions of spectrum are not at all time and in all locations used by PUs, as pointed out by measurements conducted by the Federal Communication Commission. This leads to an inefficient utilization of these bands, creating however an opportunity for unlicensed (secondary) users (SUs) that wish to transmit opportunistically over those bands. To this goal, Mitola, in 1999, proposed the concept of cognitive radio (CR), a device able to sense the environment, find the portions of spectrum temporarily left free by PUs and transmit over them. The proposed work copes with the resource allocation (RA) problem for the downlink of a cognitive secondary base station (SBS) which transmits to its secondary receivers over the licensed frequencies of the PUs, keeping the interference caused to the PUs below a certain threshold called interference temperature. In order to exploit these spectrum opportunities, a CR device needs a highly flexible physical layer, allowing it to adapt its transmission parameters (e.g. modulation order, coding size, power, etc.) to the network conditions. For these reasons, the considered SBS is a bit interleaved coded OFDMA (BIC-OFDMA) system. The multi-channel OFDM(A) modulation allows an efficient transmission over wireless selective channels and the BICM scheme offers a further robustness increase against the harsh propagation conditions of wireless channels. Recently, a huge number of works has been devoted to the problem of managing resources in underlay CR, exploiting different approaches, as for instance capacity maximization, subject to interference power constraints. The Shannon capacity, however, to be achievable, requires independent Gaussian inputs across parallel subchannels and perfect Shannon code knowledge. In practice, it results unfeasible and, for this reason, this work considers an objective function entailing practical modulation and coding schemes, as M-QAM modulation and turbo codes. Moreover the presence of the ARQ retransmission mechanism at the data link layer is taken into account in order to quantify the trade-off between data rate and link reliability. To this goal, a significant figure of metric has been detected in the offered layer 3 data rate, or goodput (GP) for short. Considering the scenario described above, the RA problem at the SBS results in finding how to jointly allocate bits and subcarriers to each secondary user in order to maximize the overall goodput realized by the system. Unfortunately, this problem has a NP-hard complexity. Thus, after a proper analysis of the goodput function, we tackled the RA problem resorting to the Ant Colony Optimization framework, which offers a valid set of tools to solve this kind of combinatorial optimization problems. The proposed algorithm, named max-sum goodput (MSG), is compared with other pragmatic algorithms bearing the same complexity and simulation results show how the MSG can remarkably boost GP performance.