IJCA, Vol. 16, No. 3, Sept. 2009 ISCA Copyright© 2009 121 An Adaptive Quality of Service Based Power Management Algorithm in Wireless Transmission Ahmed H. Salem * Hood college, Frederick, MD 21701 Anup Kumar † University of Louisville, Louisville, KY 40292 Adel S. Elmaghraby ‡ University of Louisville, Louisville, KY 40292 Abstract The increasing technological developments in the area of mobile communications enable users to gain access to a wide variety of services. These developments have led researchers to address the issue of Quality of Service (QoS) based connec- tivity for mobile network users. It is important to design a control algorithm and mechanism that can quickly determine and manage the minimum required signal power that achieves a certain data rate to guarantee the required QoS. This paper, introduces a new model to determine the minimum required signal power to guarantee a specific data rate for a user. This approach uses a genetic algorithm to estimate the shadow and fading coefficients that correspond to the maximum interference at different points in the base station’s coverage domain. The estimated parameters at different points in the domain of the base station indicate that the proposed model is accurate, effective and usable in real-time applications. Key Words: Quality of Service (QoS), mobility, genetic algorithm, power management, interference coefficients, power optimization, mobile network. 1 Introduction Today’s world of computing and business mobility is characterized by the increased use of mobile computers, PDA’s, notebooks, and cellular phones. The existing wireless networks use different algorithms to control power capacity and balancing Quality of Service (QoS) over the networks [1- 4, 7, 10, 16, 20, 23]. These algorithms estimate the power and interference ratios by using network management tools to control channel assignment and segmentation. The algorithms proposed for power distribution and energy consumption in [1, 11, 15-17, 23], require a priori knowledge of the path gains on all the radio links. In [15-17, 23] distributed power controlled * Department of Computer Science. E-mail: salem@hood.edu. † Department of Computer Engineering and Computer Science. E- mail: AK@louisville.edu. ‡ Department of Computer Engineering and Computer Science. E- mail: Adel@louisville.edu. algorithms were proposed that use only the measured C/I ratio on the active links (instead of the path gains) to adjust power and achieve C/I balancing, where (C) is the signal power of the carrier and (I) is the amount of interference affecting the power signal. The models in [15-17, 23] also incorporate peak signal power constraints that affect other nearby receivers. In power management research represented in [15-17, 23] the optimal transmitted power level is established by a set of QoS constraints. The functional form of optimal transmitted power is indicated by some function of signal to interference ratio, which does not take fading into account, and by the levels of interference in the channel. Also, it is commonly assumed that the interference is independent of the signal power at the transmitter. In these approaches, fading is not considered, and it is assumed that an infinite stream of data is waiting to be transmitted. Researchers in these cases assume an intelligent transmitter, one somehow informed regarding the level of interference at the receiver end. Finally, most power research considers basic multiple access protocols from an energy-optimality standpoint for solutions [9, 15-17, 20, 23]. This paper proposes an adaptive power control model that adjusts transmitted power based on the interference level surrounding a mobile device. This interference level could differ from one user to another. This model introduces a segmentation-based approach for calculating the interference coefficients ( path loss coefficient, and  shadowing coefficient that causes fading), for a set of users in a given coverage region. The model uses a genetic algorithm to calculate and optimize for the worst possible interference which affects the covered domain, with respect to fading and noise ratios. The proposed power optimization model is applicable to different mobile network architectures such as TDMA, CDMA, W-CDMA, GSM, EGSM and 3GSM. The key contributions of the paper include:  Integrating data rate, normalized interference and signal fading through an enhanced power transmission model.  Developing a framework for estimating gain and fading coefficients to calculate the worst possible interference at different points in the domain.  Creating an adaptive framework for guaranteed delivery of a user’s specific data rate.