Sensor Swarm query filtering: Heightened attack resilience for broadcast on-demand services

In broadcast systems a central server collects client queries for a given set of data items and schedules the next broadcast accordingly. Since anonymity has been shown to promote user participation, the client queries do not contain a client ID. However, this choice leaves ample room for service misuse. This paper identifies a number of prospective attacks and quantifies their impact on the performance of the system. A novel query filtering mechanism named Sensor Swarm is introduced to tackle these attacks. Extensive simulations show that the proposed scheme can ensure smooth system operation even under heavy attacks. Thus, a broadcast system that employs the Sensor Swarm filtering scheme can sustain excessive client misbehavior before resolving to Turing tests or compromising user anonymity.