AMEND - An Algorithm for Mitigating ENvironmental Degradations in heterogeneous networks

Many existing algorithms manage network handover using static thresholds or weights applied to performance metrics. Such approaches are performance limited as they require knowledge of prior network performance. Performance metric thresholds and weights are often preconfigured based on the experience of network personnel. Static weightings are often configured for ideal network scenarios and are not able to adapt to changing environmental conditions. Previous studies have illustrated how the combination of foliage and weather can introduce interference at the receiver. This paper proposes an Algorithm for Mitigating ENvironmental Degradations. (AMEND). AMEND is a pluggable directed feed-forward neural network designed for vehicular environments. The handover decisions implemented by AMEND are based on predicted weather conditions, historic network performance and dynamic performance characteristics. Results illustrate that in varying weather conditions AMEND has improved overall performance over existing approaches. In poor weather conditions, implementation of AMEND has led to a performance improvement of over 500% in comparison to existing approaches.