Atsuko Ikegami

Optimal Data Allocation for Keeping Fairness of Online Game

In our previous work, we proposed a load distribution technique and dynamic data reallocation method for web-based MORPG system. This approach could reduce the average server latency for game players. However, each latency widely varied, which leads unfair situation for game players. In this paper, we deal with data allocation with taking into account the fairness as a combinational optimization problem. We develop an optimization model whose objective function is to minimize the difference among latency value. We exactly solve this problem and compare the results with our previous data allocation method.

Optimal data allocation and fairness for online games

In our previous work, we proposed a load distribution technique and a dynamic data reallocation method for web-based Multiplayer Online Role-Playing Game MORPG systems. Our proposed approach reduced the average server latency for game players. However, latency values varied widely resulting in some game players gaining an advantage over others. In this paper, we deal with data allocation and consider latency fairness in the form of a combinational optimisation problem. We develop an optimisation model in which the objective function minimises the differences in latency values and we solve instances from this model using an exact algorithm. Our model improved latency fairness and more stable server loads than the previous data reallocation method. Furthermore, it provides more stable gameplay using ten web servers instead of five.

A parallel branch-and-bound algorithm for integer linear programming and its implementation on a distributed multiprocessor

This paper describes a parallel branch-and-bound algorithm for general integer linear programming problems and its implementation on a distributed memory multiprocessor nCUBE2. With a branch-and-bound algorithm, the amount of computation on each search tree node varies, and in general, is large. This has led to development of an effective communication strategy for the algorithm presented herein. Its implementation and performance evaluation also are described. The results show that the more difficult the problem instances become in terms of time to solve on a sequential computer, the more efficient the algorithm presented here proves to be.