A new max-min fairness definition to neutralize malicious users

Enforcement of max-min fairness encourages end users to implement adaptive end-to-end flow control. This pushes back the responsibility of controlling congestion to users and guarantees reliable and efficient network operation. Unfortunately, there could be malicious users who are indifferent to congestion and intentionally cause performance degradation to other users. The commonly used max-min fairness definition might allocate more resources to malicious users than they deserve and hence cause network inefficiency. This paper proposes a modified max-min definition according to which strict priority is given to cooperative users over malicious ones. We demonstrate that implementation of such two-priority fairness in local network bottlenecks will achieve controlled congestion as well as global fairness, even in the presence of malicious users. This paper motivates the need for traffic measurements to identify malicious users