As a part of the workshop on Distributed and Multiagent Planning (DMAP) at the International Conference on Automated Planning and Scheduling (ICAPS) 2015, we have organized a competition in distributed and multiagent planning. The main aims of the competition were to consolidate the planners in terms of input format; to promote development of multiagent planners both inside and outside of the multiagent research community; and to provide a proof-of-concept of a potential future multiagent planning track of the International Planning Competition (IPC). In this paper we summarize course and highlights of the competition.

Optimizing overall performance of Wireless Sensor Networks (WSNs) is important due to the limited resources available to nodes. Several aspects of this optimization problem have been studied (e.g. improving Medium Access Control (MAC) protocols, routing, energy management) mostly separately, although there is a strong inter-connection between them. In this paper an Artificial Intelligence (AI) based framework is presented to address this problem. Mixed-Observability Markov Decision Processes (MOMDPs) are used to effectively model multiple aspects of WSNs in stochastic environments including MAC in data link layer, routing in network layer, data aggregation, power management, etc. MOMDPs distinguish between full and partial observability, hence they are more efficient than other similar AI methods. The proposed framework provides global optimization of user-defined performance metrics, e.g. minimization of time delay, energy consumption and data inaccuracy. Near-optimal joint network policies are obtained via offline approximation of optimal MOMDP solutions and they are distributed among the individual nodes. Resulting node-policies place effectively no additional computational overhead on nodes in runtime. Experiments evaluate the framework by demonstrating near-optimal solutions for a small-scale WSN in detail in case of given tradeoff criteria. The proposed approach produces better joint network behavior in 5 out of 6 cases compared to other two standard methods in simulation by increasing overall network performance by more than 20% in average.