HMM-Based Admissibility Estimation Algorithm For Heterogeneous Networks

With the increase of types of business, users'personal needs, and constant pursuit of diversity, a single 3G network can not be able to meet the growing demand. WLAN, with a rapid development and popularization in recent years, has advantages in terms of its low-cost, easy to setup and other aspects of outstanding advantages. It can be choosen as a complement to 3G networks with each other and form 3G/WLAN heterogeneous networks to make better use of network resources in the hot spots. However, the integration process of heterogeneous networks still have many problems, and one of them is the access control for heterogeneous networks.In tightly coupled mode, WLAN networks can be seen as the sub-network of 3G network. so the parameters of the channels can be handled by an uniform standard. In this mode, the heterogeneous network access control is mainly based on MAHO(Mobile Assisted Handoff,MAHO). By this mechanism, users can weigh the factors such as personal preferences, tariff, network characteristics and so on, to decide which sub-network to access. The rates and other information can be obtained directly, some QoS parameters such as network delay, delay jitter, and package loss rate can be obtained by measurements, however, whether the network is available, whether the channel can support users'access requesision or not, can not be obtained directly. Among them, the admissibility can discribe whether the quality of the access channel is good or bad. It can be affected by the network QoS parameters, vary with time and associated with the admissibility of the last time. The accurate prediction of the sub-networks, can provide users with the information of heterogeneous network. Combined with other factors, it can facilitate the optimal selection process of the available sub-networks for the users.In order to obtain the characteristic feature of the admissibility, this paper proposed a network admissibility prediction model based on Hidden Markov Model (Hidden Markov Model, HMM). In this model, HMM can be measured with the network QoS parameters and the moment before the network access to the alternative network for network access of the network to forecast the network for users choose to provide the basis to judge. The HMM-based network admissibility prediction model, can use the statistics, delay, delay jitter and packet loss rate measured from the network as the observations, and the admissibility with a Markov state transition as the hidden state value. Train the parameters of HMM with the existing set of parameters by Baum-Welch algorithm, and then the sub-network's admissibility of the next moment can be obtained. Simulation results show that using this model can have a higher accuracy than the method that estimate the state only by the admissibility of the last moment. HMM network access capability It proves feasibility of the HMM-based network admissibility prediciton model, and the analysis shows that the model can make amore accurate judgments when the channel is in a relatively better or worse condition. Thus, it can be used as a pre-sentence for network selection.However, due to the complexity of the wireless channel, the network admissibility often changes with time, and sometimes there will be a mutation in the channel. To solve this problem, this paper also proposed a groping estimation model based on Mixture Hidden Markov Model (Mixture Hidden Markov Model, MHMM). The model uses not only the networks'admissibility of this moment, but also admissibility of the networks in the last moment, to estimate the admissibility of the next moment. The simulation results verify the feasibility of the model, and it is of a value of a further research.