Algorithm for adaptive error control system synthesis

In the paper an algorithm for adaptive hybrid error control system synthesis is presented. The channel is considered as a quasistationary Gilbert model under the constraint that only transitions from one stable state into two adjoining states are possible. The states are ordered according to the decreasing error probabilities. It is also assumed that the channel error statistics are known in advance. Thus, by computer simulation, the optimum codes (i.e. optimal values of the code parameters) are chosen for maximising the throughput for every channel state. During trans mission, identification of the channel states is performed continuously and the optimum code is applied. The criteria for the channel state identification are determined as suitable for block as well as convolutional codes. Channel identification and control are performed on the basis of the number of erroneous blocks within an interval of a given length. The interval lengths, as well as the corre sponding thresholds, depend on the channel state. They are optimised to achieve the minimum probability of error in channel state identification. An illustrative example of the proposed algorithm application for real channel error statistics (i.e. an experimentally obtained error sequence) is given. The approach to optimising the parameters in state identification, as presented, is more general than others found in the literature, since it can be applied to both block and convolutional codes. In the case under consideration, convolutional codes with majority logic decoding are implemented in the hybrid procedure.