Stochastic learning algorithms for adaptive modulation

Adaptive modulation has been widely studied as a means of increasing the capacity of wireless communications systems. In this paper, we present stochastic learning algorithms for the design of next-generation adaptive modulation systems. In the past, the design of adaptive modulation systems has relied on analytic and functional approximation approaches. We present a stochastic optimization algorithm for the design of adaptive modulation and coding systems. Specifically we use simultaneous perturbation stochastic approximation to adapt the parameters of the adaptive modulation system to achieve higher performance. This technique can be applied independently of channel model, error correction coding, and modulation constellation options. We show the effectiveness of this technique and discuss directions of future improvement.