Dead reckoning algorithm for multiplayer online games
Date of Issue2016-04-21
School of Computer Engineering
Multiplayer online games involve a lot of information passing between each player to render correct players’ movements and positions. In every movement a player makes, he broadcasts his movement information to other players so that other players see him moving to the new position in their display correctly. However, such broadcasting method consumes a massive amount of bandwidth, and hence, a prediction algorithm called Dead Reckoning (DR) and interest management schemes, usually implemented by using zone partitioning (ZP), were introduced to mitigate this bandwidth consumption issue by slightly reducing rendering accuracy. In this project, ZP and DR are combined to make use of the advantages of both techniques. ZP with DR was compared with an adaptive DR algorithm and ZP interest management scheme without DR to see how ZP with DR can bring improvements over existing techniques. Data for evaluation experiments were collected from a racing car simulator TORCS; each file contains the position, velocity and acceleration for all cars at a fixed poll rate. Therefore, the experiment only used the obtained game trace instead of running ZP with DR in game. The results from the experiment showed that the new technique, ZP with DR, generally resulted in lower bandwidth consumption compared to ZP or DR technique alone. Moreover, the higher complexity of ZP with DR compared to the other two techniques did not result in significant increase in run time. Therefore, ZP with DR took the advantages of both ZP and DR without significant compromise.
Final Year Project (FYP)
Nanyang Technological University