Elsevier

Computer Communications

Volume 31, Issue 11, 15 July 2008, Pages 2685-2698
Computer Communications

Improving end-to-end quality-of-service in online multi-player wireless gaming networks

https://doi.org/10.1016/j.comcom.2008.02.025Get rights and content

Abstract

In this paper, we present a novel, scalable multi-player gaming architecture that can incorporate mobile nodes. Next we propose an adaptive forward error correction (FEC) and rate-control technique to improve service quality in such a wireless gaming environment. We consider only last-hop wireless links in this paper. It is assumed that the game server and clients (mobile devices) can switch between different prediction levels having different data rates. We introduce a new scheme for estimating the packet loss rates due to congestion and wireless channel conditions and use this information in a cross-layer design to improve the overall service quality. The congestion packet loss probability is used to devise a simple TCP-friendly rate-control algorithm for sending downlink data packets from the game server. We also propose a novel adaptive FEC and dynamic packetization algorithm to alleviate the effects of wireless channel packet losses based on this chosen data rate. Extensive NS-2 simulation results show the efficacy of our scheme in achieving higher throughput.

Introduction

Online multi-player gaming poses an interesting application for wireless networks. Particularly, the so-called “first-person shooter” games are some of the most challenging interactive games to be implemented in a wireless network. Such games that involve one or more players taking some form of action against other players in the game, are not usually turn-based. As a result, game play is affected by latencies in the network: players with low latency can actually have an advantage over players exhibiting high latency in the network.

In wireless networks, the likelihood of large discrepancies between players compound the problem. This discrepancy results from the fact that players, particularly in cellular wireless networks, can exhibit a wide variance of latencies depending on their individual wireless channel conditions, mobile class and system load. For instance, a player near the center of a cell may experience much higher throughput than an user at the edge of a cell. Table 1 [34] shows the latency requirements for a few popular games. It can be seen that latencies in the order of ≈200 ms can significantly affect the quality of interactive games. Though there exists a wealth of solutions for online gaming in wire-lined networks, extending the same concepts to wireless networks open up a new set of issues to be addressed. The dynamic fluctuations in available bandwidth of a wireless channel is one such issue that can significantly affect the quality-of-service (QoS) requirements for the underlying access mechanism. The heavy-tailed gaming traffic along with considerably low bandwidth availability (compared to the wire-lined counterpart) calls for some innovative changes in the transport and application-layer protocols to provide better QoS guarantees for online gaming over wireless networks.

To the best of our knowledge, there exists no previous work in the literature for improving quality of gaming specially for the wireless links. Additionally, the main goals of this paper is to provide new solutions with the following properties:

  • (1)

    Adaptive to the changing network conditions;

  • (2)

    Incurs minimal overhead in deployment. Ideally, our goal is to provide solutions that require changes in the client (gaming device) and game server softwares without affecting the intermediate nodes in the network;

  • (3)

    Reduces the feedback overhead as much as possible. Our cross-layer design would require the receiver (mobile host, MH) to send feedback messages to the game server based on observed network conditions, and this feedback overhead in itself can undermine the benefits of adaptive QoS adjustment algorithms if not handled carefully.

  • (4)

    Fast with low run-time complexity. Nowadays, the trend is to estimate the network conditions at the MH because it has a better view of the network and the wireless channel in particular (for last-hop wireless topologies). Thus, the schemes should be low-complexity for the MHs to handle the processing overhead.

Additionally, the solutions should also be “access-independent, i.e., can provide a complimentary means for developing a game regardless of the underlying wireless access mechanisms (e.g., GPRS, Bluetooth, WCDMA, etc.). However, the schemes developed in this paper use information from the underlying access mechanism to better identify the network state and adapt accordingly and hence are access-dependent. Generalized schemes independent of the access mechanisms might be easier to implement but cannot provide appreciable performance gains. Also, our solutions depend on the packet loss probabilities due to congestion and wireless channel that are derived from the access information (assumed to be EGPRS in this work). To incorporate mobile gamers using different access types either these variables can be estimated based on the specific access type used, or simply set to zero if no information is available. Note that the rate control and adaptive FEC and packetization schemes that we develop are useful for the game server to determine how to throttle downstream gaming traffic to the mobile hosts. In the absence of approximate estimates of these packet loss probabilities, the game server can simply assume them to be zero resulting in no QoS improvement for the corresponding MHs.

This paper is organized as follows. Section 2 provides a brief overview of the architectures and QoS guarantee models in network gaming. Section 3 presents our new proxy-based game architecture. Section 4 proposes our end-to-end TCP-friendly rate control and adaptive FEC and packetization schemes to ensure QoS guarantees for wireless gaming. Section 5 presents the simulation results and Section 6 concludes the paper.

Section snippets

Related works and motivation

The focus of multi-player networked gaming has shifted to support massively multi-player online games (MMPOG). Griwodz Carsten [6] proposes a proxy-based architecture to separate the different styles of gaming traffic by defining levels of urgency and relevance for each traffic style to cope with scalability problems. A similar proxy-based architecture is also proposed in [2] to reduce the load on the central game server. A mathematical comparison of the average load on the central game server

Proxy-based game architecture model

We use the Sega Network Application Package (SNAP) [4] as the primary case study for the applications of QoS guarantees in a multi-user online gaming environment. SNAP is an application developers’ environment that provides for an efficient means of implementing multi-player online gaming. It recommends a distributed server architecture, which allows for widespread deployment of the gaming network without placing too much burden on the game developer. Its underlying transport protocol is UDP,

The QoS guarantee model

Our intention in this section is to devise a TCP-friendly Gaming Protocol for Wireless Internet called TFGWP. Due to network congestion, the gaming application should be able to alter its data rate such that it can fairly share bandwidth with other applications, and also maintain a suitable packet loss level that will not significantly affect the gaming experience of the user. Also, based on the state of the wireless channel, the protocol should be able to apply proper error correction to

Performance analysis

For the performance analysis of our TFGWP protocol, we used network simulator (NS) version 2 (NS-2) package. We implemented a standard dumbbell shaped network topology to simulate the Internet traffic which is generally used to simulate performances of TCP-friendly rate-control algorithms (Fig. 9).

The senders and receivers are kept on either side of the bottleneck link. All links (excepting the bottleneck link) are sufficiently provisioned to ensure that network congestion only occurs at the

Conclusions

In this paper, we have proposed a QoS guarantee framework to support mobile gamers in multi-player networked gaming. The proposed rate-control scheme is TCP-friendly and thus should work well during network congestion by fairly sharing bandwidth with other connections. The wireless channel fluctuations are mitigated by our adaptive FEC and packetization techniques. We have also stressed on the importance of our receiver-based schemes to minimize the overhead feedback messages. The schemes

References (34)

  • M.S. Borella

    Source models of network gaming traffic

    Computer Communications

    (2000)
  • J. Farber, Network game traffic modeling, in: Proceedings of the First Workshop on Network and System Support for...
  • M. Mauve, A generic proxy system for networked computer games, in: Proceedings of the First Workshop on Network and...
  • H. Ritter, et al., Experiences using a dual wireless technology infrastructure to support ad-hoc multiplayer games, in:...
  • SNAP Communications: Client Developers’ Guide. Available from:...
  • Griwodz Carsten, State replication for multiplayer games, in: Proceedings of the First Workshop on Network and System...
  • M. Busse, B. Lamparter, M.Mauve, W. Effelsberg, Lightweight QoS-support for networked mobile gaming, in: Proceedings of...
  • J. Pellegrino, C. Dovrolis, Bandwidth requirement and state consistency in three multiplayer game architectures, in:...
  • T. Bova, T. Krivoruchka, Reliable UDP Protocol, Networking Working Group, Internet-Draft, February...
  • J. Smed, T. Kaukoranta, H. Hakonen, A Review on Networking and Multiplayer Computer Games, Technical Report 454, Turku...
  • F. Yang, Q. Zhang, W. Zhu, Y. Zhang, End-to-end TCP-friendly streaming protocol and bit allocation for scalable video...
  • S. Biaz, N. Vaidya, Discriminating congestion losses from wireless losses using interarrival times at the receiver, in:...
  • J. Rosenberg, L. Qiu, H. Schulzrinne, Integrating Packet FEC into Adaptive Voice Playout Buffer Algorithms on the...
  • C. Chen, D.J. Goodman, Theoretical Analysis of GPRS Throughput and Delay, Research Report,...
  • ETSI, GSM 04.60 Radio Link Control/Media Access Control (RLC/MAC) protocol, v6.3.0, release...
  • Armin Heindl, Approximate analysis of queueing networks with finite buffers and losses by decomposition, Report No....
  • F.P. Kelly et al.

    Stochastic Networks: Theory and Applications

    (1996)
  • Cited by (12)

    View all citing articles on Scopus
    View full text