||I. State of the art and novelty
Minimizing the communication cost associated to a control system is a highly relevant issue that has been addressed by several works in the control literature. Traditionally, the controllers in networked control systems are designed so that they minimize the number of communications or the sampling frequency [MT11]. While in multi-agent consensus, the cost related to mechanical energy is minimized, when communication is allowed under some deterministic constraints on the position of agents [FM14,BM15]. However, in practical wireless networks, the efficiency of communication depends on the wireless channel quality which typically depends on several factors like the path loss and shadowing, in addition to a stochastic term (fast fading). The novel idea of this thesis proposal is to consider real communication costs in the control problem formulation. This topic falls
under the framework of LIA, and will involve collaboration with Mounir GHOGHO from the International University of Rabat, Morocco. Some preliminary results for energy minimization in a single agent networked control system have been obtained in [VP16]. The goal of the thesis is to use an interdisciplinary approach to problems involving multi-agent control systems which are connected through a wireless network.
II. Research Plan
In many futuristic and recently emerged applications, multiple agents (agents could be drones, robots, etc.) are required to perform complex task in a joint manner, while communicating over wireless connections. For example, a set of drones scanning an area while exchanging information they have collected. Another application is control for Unmanned Aerial Vehicles (UAVs) engaged as a wireless communication relay [L13], where a set of UAVs have to maintain communication links or deliver information packets from a source to a destination.
In [LV+16], a collaborative work with Mounir GHOGHO under the framework of LIA, the problem of trajectory planning for a robot which must download a certain number of bits from an access point, and then travel to a specified destination was studied. An optimal trajectory for the robot was found under some assumptions on the wireless channel. The goal of this Ph.D. thesis is to extend the problem treated in [LV+16], to the framework of multiple agents. We shall start with the simplest case of this extension that considers the problem in which two robots must exchange some information and then travel to their respective destinations. This case will be extended further by considering communication objectives that can only be satisfied by the collaborative effort of multiple robots.Precisely, the robots have to be controlled (in a distributed manner) so that they relay some information from
a sequence of source points A1 , A2 , …, AN to a sequence of destination points B1 , B2 , …, BN . Tools from multi-agent consensus, game theory, MDP, etc will be applied to solve various aspects of this problem.
[MT11] M. Mazo and P. Tabuada. “Decentralized event-triggered control over wireless sensor/actuator networks”. IEEE Transactions on Automatic Control, 56(10), 2456-2461 (2011).
[FM14] M. Fiacchini, I.-C. Morãrescu, “Convex conditions on decentralized control for graph topology preservation”, IEEE Transaction on Automatic Control, Vol. 59, No.6, 1640-1645, 2014.
[BM15] L. Busoniu, I.-C. Morãrescu, “Topology-preserving flocking of nonlinear agents using optimistic planning”, Control Theory and Technology, Vol. 13, No. 1, pp. 70-81, 2015.
[LV+16] D.B. Licea, V.S. Varma, S. Lasaulce, J. Daafouz and M. Ghogho, "Trajectory planning for energy-efficient vehicles with communications constraints", IEEE WINCOM, Fez, Morocco, 2016. (invited paper)
[VP16] V.S. Varma and R. Postoyan, "Energy efficient time-triggered control over wireless sensor/actuator networks", IEEE CDC, Las Vegas: U.S.A., 2016.
[L13] S. Levy, "How Google Will Use High-Flying Balloons to Deliver Internet to the Hinterlands" . Wired, June, 2013.