Abstract
Adaptive communication is among the hottest areas of research in almost all types of modern communication systems like cellular, Wi-Fi and broadcast systems. In this technique, various radio parameters like modulation symbol, code rate and transmit power etc. are customized according to the erratic channel state information (CSI) on the link. By optimal link utilization, it is proven that the technique is promising in terms of enhanced data rates and quality of service (QoS) even in poor channel conditions. Digital video broadcast-second generation (DVB-S2) has a built-in support for adaptive coding and modulation. Researchers have employed soft-computing and evolutionary algorithms to find the appropriate MODCOD (modulation/code) and power vector for the next transmission interval to combat channel hostilities and to fulfil the QoS demand. However, such algorithms take significant time to tapper-off, hence the solution may not be feasible for the real-time environments. To overcome this issue, in this paper, a real time, dynamic MODCOD and power allocation technique using a Gaussian radial basis function neural network (GRBF-NN), is proposed. Once sufficiently trained, the GRBF-NN can suggest the appropriate MODCOD and power vector for the next transmission interval based on the CSI and QoS demands per TV channel. The example data is generated by a fuzzy rule based system and differential evolution algorithm to suggest MODCOD and power vector, respectively. From the simulation results it is evident that the proposed scheme is swift and promising in terms of link utilization and QoS compared to the schemes in the literature. A spectral efficiency of 6 bits/s/Hz is achieved compared to previous approaches with 4.5 bits/s/Hz and 4.2 bits/s/Hz, respectively.