Speeding up the adaptation process in adaptive wireless push systems by applying spline interpolation technique

In wireless push systems, the server schedules the broadcasts of its information items aiming at satisfying the clients' preferences efficiently. Latest research efforts have proposed adaptive push systems, enhanced with a learning automaton, in which the server has the ability to update its estimated item demand probability vector. This vector indicates the level of the items' desirability. Even though the adaptive push systems are capable of operating in dynamic environments, where the item demand probability distribution changes periodically, the time that the learning automaton needs to adapt its estimated probability vector to a new demand probability distribution leads to degradation of the system's performance. This work addresses this problem, by applying the spline interpolation method to produce an estimation of the changed desirability immediately after this change takes place. A set of indicative feedback samples are collected by the server and the new item demand probability distribution function is approximated, providing the learning automaton with estimated item probabilities, as initial probabilities. Extensive simulation results indicate the superiority of the proposed scheme, in terms of mean response time.