[论文解读] Resource-Competitive Communication

Valerie King, Jared Saia+1 more|arXiv (Cornell University)|Feb 29, 2012

Security in Wireless Sensor Networks参考文献 77被引用 4

一句话总结

本文提出了资源竞争性通信（resource-competitive communication），一种用于分析对抗性信息交换的框架，其中爱丽丝和鲍勃在对手卡罗尔阻断信道的情况下仍能传输消息。该研究证明，即使B（卡罗尔的阻断努力）未知，且卡罗尔完全了解他们的策略，爱丽丝和鲍勃各自在期望情况下仅需O(B^{0.62} + 1)次传输时隙即可成功，表明在资源受限条件下，通信成本低于阻断。

ABSTRACT

Consider the general scenario where Alice wishes to transmit a message m to Bob. These two players share a communication channel; however, there exists an adversary, Carol, who aims to prevent the transmission of m by blocking this channel. There are costs to send, receive or block m on the channel, and we ask: How much do Alice and Bob need to spend relative to the adversary Carol in order to guarantee transmission of m? This problem abstracts many types of conflict in information networks including: jamming attacks in wireless sensor networks (WSNs) and distributed denial-of-service (DDoS) attacks on the Internet, where the costs to Alice, Bob and Carol represent an expenditure of energy or network resources. The problem allows us to quantitatively analyze the economics of information exchange in an adversarial setting and ask: Is communication cheaper than censoring communication? We answer this question in the affirmative. Specifically, in a time-slotted network with constant costs to send, receive and block m in a slot, if Carol spends a total of B slots trying to block m, then both Alice and Bob must be active for only O(B ' 1 + 1) = O(B .62 + 1) slots in expectation to transmit m, where ϕ = (1 + √ 5)/2 is the golden ratio. Surprisingly, this result holds even if (1) B is unknownto either player; (2) Carol knows the algorithms of both players, but not their random bits; and (3) Carol can attack using total knowledge of past actions of both players. In the spirit of competitive analysis, approximation guara ntees, and game-theoretic treatments, our approach represents another notion of relative performance that we call resource competitiveness. This new metric measures the worst-case performance of an algorithm relative to any adversarial strategy and pertains to scenarios where all network devices are resource-constrained. Here, we apply the resource-competitive results above to two concrete problems. First, we consider jamming attacks in WSNs and address the fundamental task of propagating m from a single device to all others in the presence of faults. Second, we examine how to mitigate application-level DDoS attacks in a wired client-server scenario.

研究动机与目标

• 分析在各方均消耗资源的对抗性网络中，通信与阻断的相对成本。
• 将现实世界威胁（如WSN干扰和DDoS攻击）建模为发送方、接收方与对手之间的资源受限博弈。
• 提出一种新的性能度量——资源竞争力，用于评估最坏情况下算法性能相对于任何对抗策略的表现。
• 为在时间分槽网络中对抗性阻断下的可靠消息传输，提供可证明高效的协议。
• 解决在无线传感器网络和有线客户端-服务器系统中资源受限条件下的实际部署问题。

提出的方法

• 将问题形式化为时间分槽博弈，其中发送、接收和阻断各消耗一个单位资源。
• 引入资源竞争力的概念，衡量最坏情况性能相对于任何对抗策略的表现。
• 设计爱丽丝和鲍勃的随机化协议，使其能够适应未知的阻断努力B，并在性能界限中使用黄金分割率φ = (1 + √5)/2。
• 证明爱丽丝和鲍勃各自仅需O(B^{ϕ−1} + 1) = O(B^{0.62} + 1)个期望活跃时隙即可传输消息。
• 确保对对抗性知识的鲁棒性：卡罗尔虽知晓算法但不知晓随机位，且可利用全部历史动作记录。
• 将理论框架应用于两个具体系统：在故障条件下用于广播的WSN，以及用于抵御DDoS攻击的客户端-服务器系统。

实验结果

研究问题

• RQ1 在资源受限的对抗性网络中，通信是否本质上比阻断更便宜？
• RQ2 当对手的阻断努力B对发送方和接收方未知时，能否保证可靠的消息传输？
• RQ3 通信协议的性能相对于对手总阻断成本B如何扩展？
• RQ4 能否构建出即使对手完全知晓协议逻辑但不知晓其随机性的资源竞争性协议，且仍保持有效性？
• RQ5 资源竞争性通信在缓解WSN干扰和客户端-服务器系统中的DDoS攻击方面具有哪些实际影响？

主要发现

• 即使B未知，爱丽丝和鲍勃各自仅需O(B^{0.62} + 1)个期望活跃时隙即可传输消息。
• 以ϕ = (1 + √5)/2定义的界限O(B^{ϕ−1} + 1)是紧致的，并在最坏情况对抗策略下依然成立。
• 即使对手卡罗尔知晓爱丽丝和鲍勃的算法并观察到所有过往动作，协议仍保持有效。
• 该结果在卡罗尔完全知晓玩家策略的前提下依然成立，只要其知识不包括其随机位。
• 该框架成功适用于无线传感器网络（用于故障条件下的广播）和有线客户端-服务器系统（用于缓解DDoS攻击）。
• 本文确立了在资源受限的对抗性环境中，通信本质上比阻断更便宜，从而正面回答了核心问题。