(DPMLA)-weighted dynamic characteristics and predictable movement learning algorithm to improve video streaming in heterogeneous environments

Abstract

Mobile video is a key driver in the growth of mobile data. In heterogeneous networking environments, multimedia sessions are particularly vulnerable to varying network capabilities of underlying networks. This paper proposes a weighted dynamic and predictable based learning algorithm to improve video streaming in heterogeneous network environments (DPMLA). Current handover methods for seamless video streaming are performance limited as they do not consider how predictability movement can be used to alter the network handover decision. Research has shown that 93% of human movement is predictable. Studies also suggest that end user movement can be reliably predicted using mobile telecom services. The DPMLA algorithm considers both the dynamic performance of the network (Received Signal Strength (RSS), delay, loss) with a measure of the predictability of end user movement. Results illustrate that the DPMLA algorithm optimizes network selection and improves overall video streaming performance.