A Framework for Realistic and Systematic Multicast Performance Evaluation

Previous multicast research often makes commonly accepted but unverifed assumptions on network topologies and group member distribution in simulation studies. In this paper, we propose a framework to systematically evaluate multicast performance for different protocols. We identify a series of metrics, and carry out extensive simulation studies on these metrics with different topological models and group member distributions for three case studies. Our simulation results indicate that realistic topology and group membership models are crucial to accurate multicast performance evaluation. These results can provide guidance for multicast researchers to perform realistic simulations, and facilitate the design and development of multicast protocols.

Punishing the Innocent along with the Guilty: The Economics of Individual versus Group Punishment

Standard models of law enforcement involve the apprehension and punishment of a single suspect, but in many contexts, punishment is actually imposed on an entire group known to contain the offender. The advantages of .group punishment. are that the offender is punished with certainty and detection costs are saved. The disadvantage is that innocent individuals are punished. We compare individual and group punishment when social welfare depends on fairness, and when it depends on deterrence. We show that group punishment may dominate in the former case if the detection technology is ineffective but never in the latter case. We discuss our results in the context of several examples.