Novel Load Balancing Algorithms Ensuring Uniform Packet Loss Probabilities for WSN

In this paper, we develop optimal scheduling mechanisms for packet forwarding in Wireless Sensor Network, where clusterheads are gathering information with a predefined Quality of Service. The objective is to ensure balanced energy consumption and to minimize the packet loss probability, subject to time constraints (i.e. different nodes must send all their packets within a given time interval). Novel solutions of scheduling are developed by combinatorial optimization, and by quadratic programming methods. In our approach, the scheduling of packet forwarding is broken down to a discrete quadratic optimization problem and the optimum is sought by a Hopfield Neural Network yielding the solution in polynomial time. The scheduling provided by the Hopfield Neural Network indeed guarantees uniform packet loss probabilities for all the nodes and saves the energy of the clusterheads. In this way, the longevity of the network can be increased.