A Vertical Handoff Algorithm Based On Fuzzy Neural Network Optimized By Genetic Algorithm

In the development of mobile communication technology, various new emerging access technologies, from 2G to 4G, from GSM to TD-LTE in the future, provide users with ubiquitous access to networks and a wide variety of businesses. Future wireless communication systems will be a seamless integration of heterogeneous wireless network based on all-IP technology. Therefore how to use heterogeneous wireless networks reliably and effectively to serve the users is one of the key issues, so vertical handoff technology will undoubtedly become a hot topic.Through the summary of previous studies, we found that the current traffic balance of heterogeneous network did not consider self-similarity of IP packets. Aiming at solve the problem that traditional network traffic models, such as Markov, Poisson is no longer applicable to describe the traffic self-similarity in future network, the paper innovatively introduced a self-similarity model named Gamma Poisson distribution to the study of vertical handoff. This model allows the simulation of network performance much closer to the actual network and has instructional meaning for he design of the network.Based on self-similar traffic sources, the threshold decision method was adopted to reduce call blocking possibility (CBP) and handoff call dropping possibility (HCDP) of the system, the proposed vertical handoff algorithm for heterogeneous networks has the pre-decision mechanism, which can reduce the number of users involved in handoff and save the system overhead. In the same time, the algorithm takes into account the signal strength parameters and sojourn time. Simulation results also show that CBP and HCDP from GPM are higher than that from Poisson model, because the service from GPM has long-time correlation. Poisson model evaluates the network performance optimistically, however, self-similarity does exist in network, so it should be considered in the design and analysis of the system.Based on the traditional threshold decision method considering self-similarity, a fuzzy neural network model optimized by generic algorithm was used in handoff decision. The model has the advantage of vertical handoff can consider more factors, such as signal strength, terminal speed, network load situation and operational speed, etc, and combine several advantages of the genetic algorithm, fuzzy logic and neural network artificial intelligence algorithms. Fuzzy neural network has a large dependence on the initial value, and easily falling into local minimum solution. The advantages of genetic algorithm can just make up for the above problem. Simulation results show that the proposed genetic algorithm based fuzzy neural network model algorithm is simple, has fast convergence speed and can reduce call blocking possibility and handoff call dropping possibility of the system to improve system performance. Comparing with the threshold decision method, network performance can be greatly improved by GA-FNN algorithm. This is mainly because the method is more flexible in the implementation process, also, various parameters can be simultaneously considered.