A RIO-like technique for interactivity loss-avoidance in fast-paced multiplayer online games

Abstract

The astonishing increase in the spread of the Internet has given rise to a globally connected community proficient at deploying online games for a large number of participants geographically located very far from each other. However, online games are characterized by more stringent requirements than traditional distributed applications deployed over the Internet can fulfill. Indeed, one of the key factors in determining the success of an online game is its ability to rapidly deliver events to the various game servers that maintain the state of the game over the network. We have already demonstrated [Palazzi et al. 2004] that in this context adapting RED (random early detection) techniques, borrowed from queuing management, can improve the global responsiveness of a game. However, this solution may not be sufficient for a specific class of online games. We deem that fast-paced multiplayer online games (such as shoot 'em ups, for example) in which participants have to behave frenetically, must guarantee a very high degree of interactivity, even at the cost of partially sacrificing the consistency of the game state. In this case having only a partially consistent view of the game state will not affect a player's amusement as much as delaying action-processing activity will. Hence we explore the possibility of applying a RIO-based (RED with in and out) algorithm to manage game delivery to the various game servers, in order to improve the degree of interactivity for fast-paced online games. Preliminary experimental results confirm the viability of our approach.

A survey on interactive games over mobile networks

“…Even if some synchronization burden is required to ensure the global consistency of the game state, this requirement is made easier, thanks to the lower number of involved GSSs. Moreover, as already mentioned, the synchronization mechanism among GSSs could smartly exploit the semantics of the game to discard few game packets in order to avoid interactivity loss when intense network traffic or computational load is slowing down the system [45,47]. For the sake of clarity, in the rest of the paper, we will refer to this synchronization mechanism, which is able to discard game events, as Fast Synchronization (FS).…”

Picasso

“…Note that contrary to perception, guaranteeing a consistent video quality does not imply a consistent data rate, due to variations in the video frames the actual data rate required to maintain a fixed video quality varies over time. Finally for the gameplay controller group, we assume that short packets are generated at random intervals with an average period of 50 milliseconds between them [21], which is slightly lower than the median human response times of 200 milliseconds. Slice Allocation: A natural question is how are slices allocated for Picasso as well as the multiple radio approach.…”

Optimizing consistency by maximizing bandwidth usage in distributed interactive applications

“…The approach detailed in this work operates by dynamically adjusting transport level reliability requirements of a DIA to minimise the volume of data transmission during periods of congestion. Another variable reliability approach is the Interactivity Loss Avoidance (ILA) mechanism [Palazzi et al 2005]. This mechanism operates on the premise that fast-paced games can tolerate a certain degree of loss without affecting user experience.…”