Markov-Decision-Based Call Admission Control with State Information of Adjacent Cells in Road-based Cellular Networks

• Main Authors: Hsien-Chung Tsao, 曹賢宗
• Other Authors: Wei-kuo Liao
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/97409882091538803809

碩士 === 國立交通大學 === 電信工程系所 === 96 === We study the admission control problem in cellular network when taking the neighboring state information into consideration with FA Strategy. For each cell, under certain assumptions, we model the system by Markov chain with two-dimensional states where the first dimension represents the base cell’s 50 states and the second dimension stands for the adjacent cells’ 300 states. As a result, the model becomes a two-dimensional Markov chain with 15000 states in total. The problem of minimizing a linear objective function of new call blocking and handoff call dropping probabilities can then be formulated as a Markov Decision Process. However, the enormous number of states makes the inverse of the transition probability matrix (which is of size 2.25×108) computation-prohibitive and thus complicates the application of policy iteration method in the context of Markov Decision Process to solve our problem. To attack such, we use the state aggregation method where we group those states which basically are few steps reachable from each other. After doing so, our model turns into involving only 66 states in total and solvable by the policy-iteration method. Finally, we show that our policy can be easily derived and has lower average cost than the well-known Guard Channel policy. However, we find that BDCL strategy outperforms FA strategy. Therefore, we modify the above MDP model by adding One-Step policy to it in order to get better efficiency.