Detection of communication signals using stochastic quantization

We consider the problem of receiving a communication signal using a stochastic signal converter with uncertain quantization levels. Deeply scaled comparator circuits may offer advantages in size and power, but can exhibit random variations in their parameters. We can take advantage of these devices by incorporating the uncertainty of the devices into the system model used for inference. In this paper, we develop a statistical model for such a converter and show how it can be applied to estimate sequences of symbols. We demonstrate the performance of the converter in a simulated communication receiver.