Research On Wireless Network Resource Allocation Based On Extended Duality Theory

OFDM technology is one of the core technologies of the next-generation mobilecommunication system with the characters such as high spectral efficiency andresistance to multipath fading and so on. Power allocation and subcarrier allocationare key issues of wireless resource allocation in OFDM system, and subcarrierallocation corresponds to a discrete problem, so resource allocation problem of theOFDM system is actually a mixed integer nonlinear programming problem (MINLP).However, existed researches on MINLP are not enough and do not have a unified andefficient method, and this paper give a common method to solve this problem.Wireless Mesh Network aggregates the technology characteristics of Ad Hoc,WLAN and WBAN networks, which is one of the most optimal networking patternsin future WMAN core network. It will probably challenge3G technologies and it isone of the most potential technologies in building B3G and4G. Among researches onresource allocation for Wireless Mesh Network, time is always neglected. This paperrepresents a resource optimization method in a typical Mesh Network withconsidering the time allocation.The main contents and innovative points in this paper include the followingaspects:1?Respecting to the nonlinear programming (NLP) problems, we introduceimproved augmented Lagrangian function. Based on it, we explain extended dualitytheory and related algorithms in detail, then design and realize an extended dualityalgorithm which can be used to solve NLP problems. Combining with annealingalgorithm, we improve the algorithm and present an extended duality algorithm whichcould solve MINLP problems. We do simulations with several examples, verify theconvergence and advantage of the designed algorithm. Finally, the2?In this paper, we also construct power allocation model for the cellularOFDMA system, which optimization goal is to maximize network utility undersatisfying the base station power limiting condition and subcarriers constraints. Thismodel includes two types of resources (power and subcarriers) allocation, which is a MINLP problem. After model transformation, we can get an equivalent continuousNLP problem model. Using the extended dual Lagrangian function, we get theextended dual problem of the MINLP model. Considering the power constraint andsubcarriers allocation constraints, we analysis penalty factors update mechanismbased on the extended dual theory. Finally, MATLAB simulation is used to check theconvergence of the algorithm. Compared to the NLP problem model result solved bygenetic algorithm (GA), the simulation also verifies that our method is much moreefficient.3? This paper constructs a system model for the typical Mesh networkconsidering the time allocation, and designs a network resource allocation schemebased on the extended dual theory. Thinking of the time slots resource, the paperestablishes the extended dual model, and then uses power constraint optimization andrate constraint optimization methods to achieve a joint optimization of power,frenquency and time, which aims to maximize the total utility of the Mesh network.Finaly, MATLAB simulation results show the superiority of the presented methods.