|dc.description.abstract||This dissertation studies control and optimization approaches to obtain energy-efficient and reliable routing schemes for battery-powered systems in network settings.
First, incorporating a non-ideal battery model, the lifetime maximization problem for static wireless sensor networks is investigated. Adopting an optimal control approach, it is shown that there exists a time-invariant optimal routing vector in a fixed topology network. Furthermore, under very mild conditions, this optimal policy is robust with respect to the battery model used. Then, the lifetime maximization problem is investigated for networks with a mobile source node. Redefining the network lifetime, two versions of the problem are studied: when there exist no prior knowledge about the source node’s motion dynamics vs. when source node’s trajectory is known in advance. For both cases, problems are formulated in the optimal control framework. For the former, the solution can be reduced to a sequence of nonlinear programming problems solved on line as the source node trajectory evolves. For the latter, an explicit off-line numerical solution is required.
Second, the problem of routing for vehicles with limited energy through a network
with inhomogeneous charging nodes is studied. The goal is to minimize the total elapsed time, including traveling and recharging time, for vehicles to reach their destinations. Adopting a game-theoretic approach, the problem is investigated from two different points of view: user-centric vs. system-centric. The former is first formulated as a mixed integer nonlinear programming problem. Then, by exploiting properties of an optimal solution, it is reduced to a lower dimensionality problem. For the latter, grouping vehicles into subflows and including the traffic congestion effects, a system-wide optimization problem is defined. Both problems are studied in a dynamic programming framework as well.
Finally, the thesis quantifies the Price Of Anarchy (POA) in transportation net- works using actual traffic data. The goal is to compare the network performance under user-optimal vs. system-optimal policies. First, user equilibria flows and origin- destination demands are estimated for the Eastern Massachusetts transportation net- work using speed and capacity datasets. Then, obtaining socially-optimal flows by solving a system-centric problem, the POA is estimated.||en_US