Broadcast Audience Estimation

We investigate the estimation of the size of a broadcast audience. We give the Maximum Likelihood estimators for different scenarios. Using Poisson approximation we obtain upper and lower bounds for different estimators. We were able to give closed-form expressions for the estimator and the bounds. We treat the problem of a malfunctioning feedback flow control, where the sender is overwhelmed by feedback messages. The influence of feedback implosion is modeled by the loss of information for the audience estimation. The EM - algorithm is used to obtain a reasonable estimate even under such condition of censored values. We further investigated the choice of the parameters in order to optimize estimation results.