Optimization of Handover Decisions for Multimedia Network Services Through Fuzzy Logic

Abstract

The increase of multimedia services in Telecommunication industry has made bandwidth a scarce valuable resource. A popular way to achieve higher capacity is the use of small cells that increases number of handovers as user move from a cell to another. Failure of handover process leads to a drop of Quality of Service, (QoS) making customers to be dissatisfied. Several handover algorithms have been proposed for handover decisions taking into consideration only few input parameters. Also, these algorithms do not take into consideration handover for multimedia services based on their QoS requirements and level of improvement on handover success rate. Service is a useful factor for the users and different services require respective QoS. This research aims at optimizing the process of handover through fuzzy logic method for cellular multimedia services by a combination of five input parameters to the Fuzzy Inference System (FIS). They are Base Transceiver Station, (BTS) traffic load, Mobile station, (MS) velocity, signal quality, signal level and the available bandwidth. In this research, the threshold for handover for the three forms of traffic (voice traffic, video traffic and web traffic) are calculated separately with priority being given to the voice call as compared to the web traffic since web traffic can tolerate some delay but voice traffic cannot tolerate any delay. Handover process is optimized by increasing the number of input parameters to the Fuzzy Inference System, (FIS). The system calculates the hand the obtained results over thresholds for the three multimedia services based on their quality of service requirements. From the obtained results voice traffic due to its stringent quality of service requirements has the highest threshold values. Voice traffic threshold values are 0.885, 0.638 and 0.337 when input parameters are worst, average and excellent respectively while web traffic the threshold values are 0.629, 0.156 and 0.110 when input parameters are worst, average and excellent respectively. Based on these threshold values, handover decision is executed. This algorithm ensures smooth and efficient handovers are executed. Many traditional handover algorithms such as Fuzzy logic based and hysteresis have short- comings. The fuzzy handover algorithm is not optimized thus needs attention from human experts. This work has analyzed the impact of using the adaptive neuro fuzzy inference system for the handover decision making. The xv results from the different simulations have shown that, need to handoff varies depending on number of inputs to the Adaptive Neuro Fuzzy Inference System, (ANFIS). The outputs are 0.334, 0.42, 0.561, 0.607 and 0.714 when the number of inputs to the ANFIS are one, two, three, four and five respectively. As the number of inputs is increased up to five, the handover decision is optimized. The data used in training the ANFIS was obtained from the developed fuzzy logic system and safaricom LTD, Kenya.