Improving Dependability of Networks with Penalty and Revocation Mechanisms

Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Dmitriy Kuptsov Name of the doctoral dissertation Improving Dependability of Networks with Penalty and Revocation Mechanisms Publisher School of Science Unit Department of Computer Science and Engineering Series Aalto University publication series DOCTORAL DISSERTATIONS 185/2014 Field of research Telecommunications Software Manuscript submitted 17 December 2013 Date of the defence 5 December 2014 Permission to publish granted (date) 31 January 2014 Language English Monograph Article dissertation (summary + original articles) Abstract Both malicious and non-malicious faults can dismantle computer networks. Thus, mitigating faults at various layers is essential in ensuring efficient and fair network resource utilization. In this thesis we take a step in this direction and study several ways to deal with faults by means of penalties and revocation mechanisms in networks that are lacking a centralized coordination point, either because of their scale or design.Both malicious and non-malicious faults can dismantle computer networks. Thus, mitigating faults at various layers is essential in ensuring efficient and fair network resource utilization. In this thesis we take a step in this direction and study several ways to deal with faults by means of penalties and revocation mechanisms in networks that are lacking a centralized coordination point, either because of their scale or design. Compromised nodes can pose a serious threat to infrastructure, end-hosts and services. Such malicious elements can undermine the availability and fairness of networked systems. To deal with such nodes, we design and analyze protocols enabling their removal from the network in a fast and a secure way. We design these protocols for two different environments. In the former setting, we assume that there are multiple, but independent trusted points in the network which coordinate other nodes in the network. In the latter, we assume that all nodes play equal roles in the network and thus need to cooperate to carry out common functionality. We analyze these solutions and discuss possible deployment scenarios. Next we turn our attention to wireless edge networks. In this context, some nodes, without being malicious, can still behave in an unfair manner. To deal with the situation, we propose several self-penalty mechanisms. We implement the proposed protocols employing a commodity hardware and conduct experiments in real-world environments. The analysis of data collected in several measurement rounds revealed improvements in terms of higher fairness and throughput. We corroborate the results with simulations and an analytic model. And finally, we discuss how to measure fairness in dynamic settings, where nodes can have heterogeneous resource demands.