**Eavesdropping a noisy radio channel by manipulation of photon bunching statistics -- theory and simulated performance** - Richard Lieu

The use of higher moments of an intensity time series of radio observations as extra information to improve the sensitivity of receivers is proposed. It is shown that a faint source of mean flux $\ep$ embedded in the time series of background noise of mean flux $s\gg\ep$ can be detected with increasing statistical significance if more higher moments of the time series are involved in the construction of one's flux measure, in accordance with a specific hierachical formulation. However, if $n$ is the order of the highest moment being employed then, for the method to be effective the larger the $n$ the better the resolution one must employ to sample the data. In this work we present analytical predictions of the eavesdropping performance, accompanied by numerical simulation of stationary and fully incoherent light of high occupation number (the radio limit) where the principal noise component is phase (or photon bunching) noise, to show that the predictions and simulations are in good agreement. We restricted our attention currently to the $n\ep \ll s$ limit of relatively faint sources. Sources that do not satisfy these conditions, cannot easily be modeled in analytical terms, and only further simulation effort can reveal the best way of exploiting the proposed ideas to observe them. Nevertheless, it is believed the recipe presented here opens the window to a different approach to radio data processing, one that might hold the key to the discovery of many faint sources of different types, and also such detailed radiation properties of a bright source as the coherence length of quasar emission.