The Research On QoS Guarantee Mechanisms Of Wireless Networks Based On Intelligent Control Theory

With the development of Mobile communication techniques and Internet networks, it has been a challenging research topic to provide informantion services for mobile subscribers at anytime and at anywhere. The future mobile communication will be IP-based core networks which realize the user access through wired links and wireless links. The expected services will inculude voice, data, vidio, E-mail, mobile multimedia and mobile Internet access. In order to support different types of services, the next generation mobile networks must have a series of methods to satisfy Quality-of-Service (QoS) requirements of the users and to optimize the resource allocation for wireless networks while maximizing the total network performance. As different kinds of traffic has different QoS requirements, such as packet error rate, transmission rate, delay limitation and radio resource demands, providing QoS guarantees for all types of services in mobile networks is an inherently challenging task and it involves multipile network protocol layers.Comparing to fixed networks, the QoS guarantees in wireless networks face two problems: (1) Due to the limtation of spectrum resource and channel bandwidth fluctuations, the utilization of radio resource and system stabilization have to be improved. (2) The physical layer problems, such as multipath fading loss, high Bit-Error-Rate, and handoff loss, can make the communication links unreliable and recede system performance. Recent years, the research interests of QoS support techniques in wireless networks focus on the careful mangement of the common radio resources in the radio access networks and the improvement reliability of transport mechanisms in the core networks.In the third generation (3G) CDMA systems, the fast power control and admission control techniques are widely used to optimize the radio resource allocation, maintain the system stabilization and satisify the QoS requements of the users. On the other hand, next generation mobile networks will be 'all-IP' networks based on the Internet Protocols in which transimission is unreliable. TCP is the most widely used transport protocol in the current Internet. TCP's performance over wireless networks, which is dominated by the congestion control mechanisms, is very important for future mobile subscribers. The main reason that influences data transimission quality (delay, jitter) and recedes network resources (bandwidth, buffer) is network congestion. It is a very complicated kernel issue to realize the radio resource management and congestion control in wireless networks effectively under QoS and fairness constraints.The intelligent control theory and techniques developed these years, includingneural network, fuzzy logic, expert control and genetic algorithms, can be used to solve complicated control problems which are unanswerable with classical control theory. People begin to present the intelligent control theoretic approach to network control mechanisms. We study the QoS guarantee mechanisms of wireless networks based on the intelligent control theory and techniques. The main contributions of this paper can be summarized as follows: (1) As power control is the most important aspect in a WCDMA systems to optimize the resource allocation for wireless networks and to satisfy QoS requirements of the users, an uplink power control algorithm using hybrid neural network techniques is present. Based on the CRC detection of wireless channel and FER target criteria in real time, the SIR target and the transmit power of mobile station are controlled by different neural network controllers. This method provides sensitive response to the different communication environments. Comparing with the conventional fixed-step approaches, this algorithm enhances control accuracy and control speed, improves the control system stability, and provides a better QoS criteria and more system capacity. (2) An intelligent call admission control algorithm using fuzzy neural network for the radio resource management of WCDMA systems is proposed. According to the QoS measures of all service types, the intelligent call admission controller (ICAC) adopts fuzzy logic to estimate new call interference and uses the neural network techniques to predict next-step existing-call interference, then making admission decision for a new call request. It consists of a fuzzy interference estimator, a neural network interference predictor, and a call admission processor. ICAC is more adaptive and attains better performance than the SIR-based CAC approaches in WCDMA systems. (3) A new TCP congestion control improvement algorithm based on bandwidth estimation is proposed. The algorithm estimates the bandwidth based on the packet arrival time carried by acknowledgement packet and uses the estimated bandwidth to update congestion window. The experiment results show that the new algorithm has higher performance compared with traditional TCP protocols such as TCP Newreno and TCP Sack when the links have high Bit-Error-Rate. The improvement algorithm is more suitable for wireless networks. (4) Traditional TCP congestion control algorithms can yield a significant throughput deterioration when applied to wireless networks without modification since the primary cause of packet loss may not be network congestion. This paper presents a congestion control mechanism for wireless networks based on expert control, which makes sure the state of TCP connection and differentiates between congestion and wireless losses in terms of the inference of network expert knowledge. According tothe cause of packet loss, different congestion control strategies are adopted. The simulation results show that this mechanism can accurately differentiate between wireless loss and congestion and improve the throughput of the network, and it is more robust than traditional congestion control mechanisms of TCP/IP. (5) Three kinds of factors will mislead up to congestion control and dominate TCP's performance in the third generation (3G) mobile networks, which are relatively high Bit-Error-Rate of radio propagation channels, handoff loss and undesired data loss causing by the unmatched rate between wired and wireless links. By deploying relay windows at the boundaries of wireless links, a TCP performance optimal control strategy is proposed, which utilizes three sliding windows mechanism to establish the TCP connection and maintains end-to-end TCP semantics. By the shielding effect of relay windows, the wireless losses can be differentiated from congestion efficiently. According to the radio resource and the feedback information of TCP connection, the relay windows can be used to control the sending rate, balance the bandwidth difference of wired and wireless links and prevent an undesired packets overflow of the buffer. The packet losses and throughput descent of handoff duration can also be reduced by the copy of the relay window information. The simulation results show that this scheme can adapt to mobile networks and improve the TCP throughput performance.