Neural network implementation of a fade countermeasure controller for a VSAT link

In this paper, an algorithm and its practical implementation to activate adaptive modulation as a fade countermeasure (FCM) is presented. Characteristics of the algorithm are derived from the propagation studies and the algorithm uses a simple four parameter model. A neural network architecture was used to implement the decision making block of the controller. The algorithm has been implemented on a fixed point DSP. An experimental set-up with an emulated K_a-band satellite link and a terrestrial return path connection has been used with previously recorded propagation data for experimental verification and performance analysis. Performance of the implemented FCM system is compared with that of fixed, non-adaptive systems. Over concatenated rain events, the adaptive system yields better throughput at or below a given BER than any fixed mode system and only marginally worse BER availability than the most robust scheme in the system. The FCM is thus essential if throughput is of importance as well as availability. The added complexity of the FCM system is not great by contemporary technology standards and is, in authors' opinion, well worth the investment.