The growth and penetration of broadband access networks to the home has fueled the growth of online games played over the Internet. As we write this article, it is 5am on a typical weekday morning and Gamespy Arcade 1 reports more than 250,000 players online playing about 75,000 games! This proliferation of online games has been matched by an equivalent growth in both the

Massive multiplayer online games (MMOGs) are today the driving factor for the development of distributed interactive applications, and they are increasing in size and complexity. Even a small MMOG supports thousands of players, the biggest support hundreds of thousands of concurrent players. Since they are typically built as strict client-server systems, they suffer from the inherent scalability problem of the architecture. Computing power and bandwidth limitations close to the server limit the possible number of players. Also, the latency of communication between players through the server will be higher than using direct communication. In the paper, we address these issues and investigate improvement options. A typical MMOG consists of a virtual world with a concept of time and space that is similar to the real world. In it, players are represented by avatars. Only subsets of these avatars interact with each other at any given time. This allows us to divide them into groups, and communication among group members becomes a multi-party communication problem. Thus, to reduce resource consumption, we compare the performance of several algorithms for group communication with the current central server approach. We use overlay multicast as the means of providing group communication, and research algorithms for creating shortest path trees, spanning trees, delay-bounded spanning trees and, more specific, applying Steiner tree heuristics. Our experimental results indicate that different approaches are useful to reduce resource consumption while achieving a good perceived quality under varying conditions, such as frequent changes in group membership and the demand for low latency. 1.

... (MMOGs) can include millions of concurrent players spread across the world. To keep these highly-interactive virtual environments online, a MMOG operator may need to provision tens of thousands of computing resources from various data centers. Faced with large resource demand variability, and with misfit resource renting policies, the current industry practice is to maintain for each game tens of self-owned data centers. In this work we investigate the dynamic resource provisioning from external data centers for MMOG operation. We introduce a novel MMOG workload model that represents the dynamics of both the player population and the player interactions. We evaluate several algorithms, including a novel neural network predictor, for predicting the resource demand. Using trace-based simulation, we evaluate the impact of the data center policies on the resource provisioning efficiency; we show that dynamic provisioning can be much more efficient than its static alternative.

With the increasing popularity of massively multi-player online games (MMOGs), developers are continually forced to deal with the conflicting requirements of supporting a large number of concurrent users, while simultaneously providing low latency. As a result, a common way of distributing load is by dividing the virtual environment into logical regions. Geographically coupled users in such regions can be distinguished by analyzing IP addresses, RTTs or similar. This paper proposes the architecture for a decentralized middleware capable of utilizing such information, with the intent of decreasing the overall latency for the majority of users in that region. The latency reduction is accomplished by migrating a game region to a server closer in locality to the users, thereby lowering the response time of remote procedure calls. Due to the characteristics of MMOGs, the middleware implements a distributed name service, made possible by activating the system from a single node in the system. 1

Massively multi-player online games (MMOGs) require low latency while supporting a large number of concurrent players, often sharing one worldwide instance of the game. As these are conflicting requirements, a common way of distributing load is by dividing the virtual environment into virtual regions. As MMOGs are world-spanning games, it is plausible to disperse these regions on geographically distributed servers. As such, we propose the use of core selection for finding an optimal server for placing a region, and support for migrating the game state to that server. The first goal relies on a set of servers and measurement of the interacting players latencies. In locating an optimal server, we anticipate a decrease in the overall latency for the majority of players. This reduction occurs by migrating the region to a server closer in proximity to the majority of players in that virtual region, thereby lowering the response time of any interaction. 1.

Large improvements in computer technology allow thousands of users to concurrently interact in a virtual game. Due to this development, the body of work analyzing game traffic has grown considerably in the recent past. However, little work has been done to examine and compare networking techniques with respect to meeting the stringent latency requirements that are common for networked games. Most interactive games need response times between 100 and 1000 ms depending on the game genre [6]. In this paper, we evaluate different techniques for delivering packets in a timely manner. In particular, we compare existing user-space middleware running on top of UDP and reliable, fair transport protocols like TCP and SCTP. In addition, we evaluate some “low latency ” extensions to TCP, SCTP and one of the middleware platforms. We present results concerning packet latency and bandwidth requirements for the different approaches. 1.

The popularity of distributed interactive applications has exploded in the last few years. For example, massive multi-player online games have become a fast growing, multi-million industry with a very high user mass supporting hundreds or thousands of concurrent players. Today, such games are usually client-server applications that use TCP for time-dependent communication. Similar multimedia applications also choose TCP frequently. Very thin data streams are sent over each of these TCP connections, which means that they consume very little bandwidth. TCP has several shortcomings with respect to the latency requirements of such thin streams because of its retransmission handling [7]. An alternative to TCP may be SCTP [13] which was developed to answer the requirements for signaling transport. SCTP has subsequently also been considered more appropriate than TCP for congestion controlled streaming of other time-dependent data. Important reasons are its maintenance of packet boundaries and partial reliability. In this paper, we evaluate the performance of the Linux SCTP implementation for thin streams. Like others before, we identify latency challenges. We also propose some enhancements for reducing the latency compared to the original Linux implementation. We argue for separate handling of thin and thick data streams in SCTP.

Latency reduction in distributed interactive applications has been studied intensively. Such applications may have stringent latency requirements and dynamic user groups. We focus on application-layer multicast with a centralized approach to the group management. The groups are organized in overlay networks that are created using graph algorithms that create a tree structure for the group. A tree has no cycles and uses a small routing table, as opposed to a connected overlay mesh. We investigate a group of spanning tree problems that are referred to as Steiner tree problems, and we have a particular focus on reducing the diameter of a tree, which is the maximum pairwise latency in a tree. In addition, we focus on reducing the time it takes to execute membership changes. In that context, we use coreselection heuristics to find well-placed client nodes, and edge-pruning algorithms to reduce the number of edges in an otherwise fully meshed overlay. Our edge-pruning algorithms strongly connect well-placed client nodes to the remaining group members, to create new and pruned group graphs. Consequently, when a tree algorithm is applied to a pruned group graph, it is manipulated into creating trees with a smaller diameter. We devised new Steiner-tree heuristics that reduced the diameter, and also proposed new edge-pruning algorithms to make the tree construction faster. These heuristics and algorithms were implemented and analyzed experimentally along with several spanning-tree and core-selection heuristics found in the literature. We found that a full-mesh of shortest paths makes it difficult for Steiner-tree heuristics to find better trees than spanning tree algorithms. The network seen from the application layer is in fact a full mesh of shortest paths. In addition, we found that faster Steiner-tree heuristics that do not explicitly optimize the diameter are able to compete with slower heuristics that do optimize it.

Abstract — With the growing popularity of wireless local area networks (WLANs) has come an increased need for effective measurements of real-world WLANs and their applications. This paper presents tools and techniques for measuring IEEE 802.11 WLANs. The techniques include details on setting up a PC as a wireless access point and building a wireless sniffer while the tools include programs for measuring link, network and application layer traffic. The tools are all open-source software available for download and the techniques all use open-source software and off-the-shelf hardware components. Together, these tools and techniques facilitate WLAN performance analysis across network layers in a flexible, accurate and cost-effective manner. To illustrate the usefulness of these tools and techniques for gathering WLAN measurements three case studies are presented: a streaming video session showing cross-layer performance; network characteristics of a wireless hand-held game; and measurements of access point queue size. Research employing these tools can yield more accurate WLAN models and more realistic evaluation of proposed WLAN changes in a network testbed. I.

Abstract — Traffic modeling is a fertile research area. This paper proposes a packet-level traffic model of traffic sources based on Hidden Markov Model. It has been developed by using real network traffic and estimating in a combined fashion Packet Size and Inter Packet Time. The effectiveness of the proposed model is evaluated by studying several traffic types with strong differences in terms of both applications/users and protocol behavior. Indeed, we applied our model to real traffic traces of Age of Mythology (a Multi Player Network Game), SMTP, and HTTP. An analytical basis and the mathematical details regarding the model are given. Results show how the proposed model captures first-order statistics, as well as temporal dynamics via auto- and cross-correlation. Also, the capability to accurately replicate the considered traffic sources is shown. Finally, preliminary results for model-based traffic prediction reveal encouraging. I.