Optimizing event distribution in publish/subscribe systems in the presence of policy-constraints and composite events

In the publish/subscribe paradigm, information is disseminated from publishers to subscribers that are interested in receiving the information. In practice, information dissemination is often restricted by policy constraints due to concerns such as security or confidentiality agreement. Meanwhile, to avoid overwhelming subscribers by the vast amount of primitive information, primitive pieces of information can be combined at so-called brokers in the network, a process called composition. Information composition provides subscribers the desirable ability to express interests in an efficiently selective way. In this paper, we formulate the min-cost event distribution problem in pub/sub systems with policy constraints and information composition. Our goal is to minimize the total cost of event transmission while satisfying policy constraints and enabling information composition. This optimization problem is shown to be NP-complete. Our simulation study shows that our heuristics work efficiently, especially in a policy-constrained system. We also find that by increasing the number of broker nodes in a pub/sub system, we are able to reduce the total cost of event delivery.