A sequential descent DCA algorithm for fast and optimum resource allocation in wireless communication systems

Hopfield neural network (HNN) has recently received increased attention as a means of solving DCA (dynamic channel allocation) problems. HNN is characterised by its parallel structure which in hardware implementation produces optimum solutions in a short time. This is in contrast to the software implementation which is slow for solving DCA problems. Software implementations are often used in simulation studies and they may also be considered for real time DCA, provided the computation delay is acceptable. This paper presents a new algorithm called sequential descent DCA (SD-DCA) which applies similar rules to the HNN counterpart but is based on a sequential refinement scheme rather than a parallel neural algorithm for solving a DCA problem. Compared to the software implementation of the neural algorithm, the proposed scheme improves performance in terms of the time required to obtain a solution.