Measuring Information Exposure Attacks on Interest Management

Abstract

Abstract-For a scalable Massively Multiplayer Online Game (MMOG), interest management (IM) is an essential component to reduce unnecessary network traffic. As Area-Of-Interest (AOI) defines each player's interests, an entity normally maintains a subscriber list of players whose AOIs cover the position of the entity. To maintain the subscriber list, players are required to send AOI updates. Unfortunately, AOI update is vulnerable to information exposure (IE) attack especially on P2P infrastructure. Sensitive information, such as player's position, can be revealed during AOI update without owner's authorization and attention. This eventually results in an unfair game. In this paper, we demonstrate that such IE attack on MMOG can help cheaters gain unauthorized benefits. Notably, we present a Monte Carlo based simulator to quantitatively measure the impact of IE attack when different IM schemes are applied. Three P2P schemes are assessed and a Client/Server scheme is also employed for comparison. In addition, we also evaluate IE attack when a group of players collude with each other to share information. Experimental data obtained from simulation are analyzed and explained. Practical suggestions are also given for choosing an IM scheme for P2P gaming.

Online classifier adaptation for cost-sensitive learning

“…Moreover, the loss function of Adaboost will be modified to consider the classification costs. Moreover, we will also investigate the application of the proposed algorithm to information security [41,43,53,42,41,52,18], bioinformatics [40,24,57], medial imaging [56,55,54], computer vision [38,23,22,36,37,19,17,46,44,45,15], reinforcement learning [27,28], cloud computing [50,51] and microprocessor reliability modeling [8,6,7,5,59,61,60].…”

“…The algorithm takes an initial classifier f (x) as an input, and operates on a iterative way. In each iteration, one new training sample is input to update the classifier adaptation function parameter, based on the updating rules in ( 15) - (17), and (18).…”

Interest management for distributed virtual environments

“…Although the official game client does not show the HP values on the screen, the player may still use third-party programs to access such information, thus gaining an unfair advantage. A quantitative analysis of information exposure attacks on interest management can be found in Hao and Cai [2012].…”