Fair overload handling using proof-of-work functions

Overload can reduce the overall performance of a computer system up to a point where the whole service might collapse. Thus, for example the load on a server must be controlled to achieve optimal system performance. Doing so some requests from clients must be rejected. We show that we can distinguish four different types of user profiles, why micropayments are no suitable solution and how a proof-of-work based system can be used to allocate the rare resources slots to those clients that generate the maximal use out of these requests. We then present the new HashCashLin function we developed in order to overcome the poor adjustment possibilities of the classic HashCash function and how such a proof-of-work can be augmented. We describe our proof-of-work auction system and finally we show the results obtained by running a simulated model of our system