科研之友
微信
新浪微博
Facebook
分享链接
摘要
The resource (bandwidth) allocation in a network is usually casted into a so-called network utility maximization (NUM) problem, which solution strategy has successfully generated distributed algorithms for congestion controlling of elastic traffic in a wire-lined network. However, for resource allocation of inelastic traffic including soft QoS (quality of service) traffic in a wireless network, this approach still faces challenges. First, it is hard for the wireless system to dynamically model the utility function of the users. Second, the utility function of soft QoS traffic is usually nonconcave, which brings the NUM optimization problem to be mathematically intractable. With deviation to the usual NUM theory, this paper proposes a novel optimization model and its algorithm to allocate bandwidth around the user's desired value to the soft QoS traffic in a wireless network. Our approach takes advantage of the basic feature of soft QoS traffic; that is, it demands a preferred amount of bandwidth but allows some flexibility during normal operation. Compared with the utility-based approaches and solutions, our approach avoids the difficulty of finding the exact utility function expression for each user by using the preference bandwidth value. This facilitates the operation of real wireless networks. The proposed model and algorithm are verified by an example, which demonstrate better performance than the NUM approach.
