Abstract
In wireless communications, packet size is one of the most significant parameters that have an impact on the quality of service (QoS). The optimal packet size depends on wireless channel condition and reception performance. However, it is difficult to determine the optimal packet size under time-varying channel in advance. Besides, performance characteristic of wireless devices is virtually unpredictable due to different hardware implementations. In this paper, we propose an approach to estimate the optimal packet size based on some of pre-measurement results which are shown later that they can be considered as a nonlinear multi-regression problem. The collected data contain the knowledge about the effect of wireless reception. Based on this information, the pseudo gradient search approach in combination with the 2-additive measures is adopted to determine the optimal search direction, build knowledge about reception performance, and accurately estimate the optimal packet size. The main contribution of this paper is the theoretical model of optimal frame size determination as well as the validation of our proposed approach through numerical analysis.