Physical layer security via secret trellis pruning

Constructions of secure channel encoders based on trellis pruning are considered in this paper. The key defines how pruning is applied on the trellis of a mother convolutional code; this results into a secret pruned trellis that legitimate users are using to perform decoding, in contrast to the eavesdroppers that employ the full mother trellis diagram. We focus on two special forms of the pruning function, and in each case we compute the expected weight enumerating function of the secret pruned code. The theoretical analysis ensures that the legitimate users achieve superior performance, in terms of word and bit error rate, than the eavesdroppers, which depends on the pruning rate. We also derive design guidelines on properties that mother encoders must have to fully exploit the proposed scheme. Simulation results also show the potential of catastrophic encoders for PHY security and yet the ability of the legitimate users to communicate reliably.