Currently, my research is focused on creating highly scalable Massively Multiplayer Online Games (MMOGs), using peer to peer architectures.
Currently, client-server architectures are employed to host MMOGS. Multiple server solutions exist, using separate servers to run copies of the game world. Other techniques exist, where distributed server architectures are employed to host a single game on a large server grid. The hardware to run such a system is very expensive and a more cost effective solution is sought. Other problems such as reduced redundancy, higher latency and lower scalability also exist.
I am investigating using peer-to-peer architectures, where each peer adds sufficient resources to the network to host itself. This technique promises to greatly reduce the costs of game hosting and maintenance, while improving responsiveness. Higher responsiveness is achieved by using direct connections, instead of communicating via the server, which is the technique currently used in MMOGs. This method will effectively half the latency, enabling games with stricter latency requirements to be implemented in MMOG form.
A P2P solution does however have many issues that have to be solved before this can become the new norm. This include network security: how does one detect and ensure that cheaters do not ruin the game for everyone. Solving the security issue is particularly tricky in a peer-to-peer context as a malicious user has access to the complete system as opposed to only the client code in a client/server architecture. This greatly simplifies the security for the client/server model.
Another issue is how to store the game state in a distributed fashion, that allows for data retrieval with low latency and high redundancy and security.
I’m developing a truly distributed P2P MMOG architecture that promises low latency and a high degree of state persistency. The focus of my work is on designing an MMOG architecture with a novel state persistency mechanism. For the remainder of the architecture, components that fit into a truly distributed design will be used. These include finely grained Interest Management techniques and hybrid game event dissemination.
Oversim Demystified [Website], Dilum Bandara, John S. Gilmore, May 2012
A Survey of State Persistency in Peer-to-Peer Massively Multiplayer Online Games [IEEEXplore], John S. Gilmore and Herman A. Engelbrecht, IEEE Transactions on Parallel and Distributed Systems, May 2012
Machine-to-Machine (M2M) Communications in Vehicular Networks [pdf], Marthinus J. Booysen, John S. Gilmore, Sherali Zeadally and Gert-Jan van Rooyen, Transactions on Internet and Information Systems, February 28, 2012
Pithos: A State Persistency Architecture for Peer-to-Peer Massively Multiuser Virtual Environments [pdf], [IEEEXplore], John S. Gilmore and Herman A. Engelbrecht, Proceedings of The 4th International Workshop on Massively Multiuser Virtual Environments (MMVE) at the IEEE International Symposium on Haptic Audio-Visual Environments and Games (HAVE 2011), 15 October, Qinhuangdao, China
Predicting Low Earth Orbit Satellite Communications Quality and Visibility over Time [pdf], John S. Gilmore and Riaan Wolhuter, Proceedings of the 12th Southern Africa Telecommunication Networks and Applications Conference (SATNAC 2009), Swaziland
A Multi-channel satellite scheduling algorithm [pdf], John S. Gilmore and Riaan Wolhuter, Proceedings of the 11th Southern Africa Telecommunication Networks and Applications Conference (SATNAC 2008), South Africa
Mapping the African Internet [pdf], John S. Gilmore, Nico F. Huysamen, Phillip Cronje, Marc C. de Klerk and Anthony E. Krzesinski, Proceedings of the 2nd IFIP International Symposium on Wireless Communications and Information Technology in Developing Countries (WCITD 2008), Pretoria, South Africa