EMULSIoN: Environment Mitigation on mULtimedia StreamIng Networks

Handover algorithms typically operate by assigning preconfigured threshold or weight values onto network performance metrics such as delay, data loss and signal strength. Such approaches are performance limited as they do not consider external factors that affect the network such as the physical environment and current weather conditions. Previous research illustrates that foliage density combined with detrimental weather conditions can have degrading effects on wireless links. The changes to these environmental factors over long vehicular-based mobile user sessions can lead to sub-optimal handover decisions and a negative impact on a user´s Quality of Experience during mobile video streaming. There is need for a handover approach that adapts to these factors and mitigates any negative effects that occur. This paper proposes a method for Environmental factor Mitigation on mULtimedia StreamIng Networks (EMULSIoN). EMULSIoN uses a perceptron artificial neural network approach to mitigate the latency and delays caused by environmental factors. Using dynamic network performance metrics and with known topographical data, the EMULSIoN directed learning approach can learn from previous user sessions to mitigate these environmental effects. EMULSIoN further uses GPS and topographical data to divide vehicular routes into small sub-areas for optimal performance in varied terrain. Results illustrate that EMULSIoN has significant video quality improvements in comparison to pre-configured weight handover strategies.