Markov Decision Process Based Multiple Codes Assignment in UMTS WCDMA Mobile Networks

• Authors: Ben-Jye Chang
• Venue: Wireless Personal Communications: An International Journal
• Year: 2007

For achieving high transmission rate in mobile multimedia communications, 3G WCDMA systems adopt the Orthogonal Variable Spreading Factor (OVSF) code tree to assign a single channelization code for each accepted connection. Based on the orthogonal characteristic of an OVSF code tree, an allocated code blocks the channelization codes that are on the descendant branches and the ancestral codes of the allocated code. Several researches have been proposed to overcome the code-blocking problem for maximizing system utilization. By using both the code assignment and reassignment mechanisms, the system utilization and code blocking can be improved. Nevertheless, the data rate of traffic classes in such single code assignment system should be powers of two of the basic rate, which is impractical and wastes the system capacity when the required rate is not powers of two of the basic rate. A good solution is to assign multiple codes to a new connection, but causes two drawbacks: high complexity of handling multiple codes and high cost from using more number of rake combiners. Consequently, there is a trade-off between waste rate and complexity of handling multiple codes assignments. In previous researches, high computation complexity of assigning multiple codes for a connection and large number of reassignment codes suppressed the advantage of reducing waste rate. Therefore, we propose an adaptive efficient partition algorithm with the Markov Decision Process (MDP) analysis approach to reduce the large number of reassignment codes while improving waste rate. There are two main motivations in the proposed approach. First, we propose an adaptive efficient partition algorithm to determine multiple codes based on the current state of the OVSF code tree for the new incoming connection. Second, after determining the multiple codes, we adopt the MDP analysis to assign the least-cost code for each determined code. Numerical results demonstrate that the proposed MDP approach yields the least number of reassignments and the least number of codes per connection while reducing waste rate significantly, as compared to other approaches.