Truthful load-aware service selection: A mechanism design method

QoS-aware service selection is a kind of classical problem in service computing and has been widely studied. A common assumption made by the majority of existing solutions is that the candidate service will report its true QoS value honestly. However, this may not be true in open Internet environments since some service providers are selfish who wants to get more profits when providing services. Instead of revealing the true cost, the provider may declare any false cost to get greater profits. In this paper, we investigate the problem of load-aware service selection involving such service providers where load is a kind of QoS parameter. We formulate the problem as finding a least response time path. According to the mechanism design theory, we design a pricing scheme which can be used to incent each provider to reveal its true load to maximize its utility. A service pricing algorithm is presented to calculate prices paid to providers in a distributed manner. Our experiment shows that the composite service performance will be degraded due to false bids declarations while the response time of the composite service is optimal when each service provider reports its true load under the control of the suggested mechanism. Our mechanism also makes service providers load-balanced in the same service class. And at the same time, the system overpayment will decrease to a lower level.