The effect of a cheater on the performance of a random-access protocol

We consider a synchronous broadcast communication channel. This channel is used by a large population of users abiding to a random-access protocol of the Capetanakis-Tsybakov-Mikhailov type, and at the same time by a cheater, with distinguished characteristics and behavior. We assume that the cheater ignores the protocol used by the other users and model his transmission attempts as a Bernoulli process. We then obtain throughput and mean delay results for both the original protocol-abiding population and the cheater. The results show that the cheater's delay is considerably lower than that of the protocol abiding population. However, there is no abrupt failure of the protocol. The stability region of the original protocol is diminished, but the decrease is almost linear with the transmission rate of the cheater. Notice that this model and analysis can serve in studying and suggesting non-obvious solutions to random-access environments with heavy users (e.g., gateways), station malfunctions, and deliberate user behavior. The obtained difference in performance can also be exploited in order to provide various degrees of priority to specific users.