What are Delay Tolerant Networks?

  • The main characteristic of Delay Tolerant Networks (DTNs) is their lack of end-to-and and of long term connectivity. This may stem from a collection of causes such as limited radio range and energy, node sparsity and mobility, jamming, external noise etc. Given this situation, the only means to propagate information is DTNs is the contact: an occasional and relatively short time period of temporary physical proximity of two devices.

  • DTNs may arise in a variety of environments and almost always include very heterogeneous devices. From urban environments to naturally unfriendly places (desert, high mountain top, the Poles) and including developing areas with little to no infrastructure, DTNs have enormous potential.

  • The DTN setting presents entirely new networking challenges, in comparison with the traditional wired network view and even with early mobile ad hoc network scenarios, which were always considered connected. In particular, it is crucial to be able to predict and use contact patterns, and to optimally employ the available resources, i.e., contact capacity, buffer space, energy in order to solve problems, old and new: routing, resource allocation, content placement etc.

Heterogeneous node mobility in DTNs

  • An aspect of paramount importance in networking research is having sound analytical models. They allow us to better understand tradeoffs, counter-intuitive behaviors that we may experimentally observe, and they can reveal unsuspected levers.

  • Initially, DTN analytical models traded accuracy for simplicity and tractability by assuming homogeneous node mobility. Bringing more realistic, heterogeneous node mobility into the picture is slowly under way, but the considerable complexity involved has, so far, hindered any generic conclusions, of the kind obtained from homogeneous mobility analyses. This is one the aims of our reserach.

Common features of DTN algorithms

  • Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

  • Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum. Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

Abstract for my thesis proposal

The growing number of mobile, networking-enabled devices (smartphones, tablets) implies that anyone can set up a network anywhere, anytime and with anybody, especially in urban environments. Ad hoc networks of this type, also known as Opportunistic Networks, form a subset of Delay Tolerant Networks (DTN). DTNs are sparse, with only partial and sporadic connectivity, in short, a very challenging communication environment. As a result, most DTN solutions to common networking problems (routing, resource allocation, content dissemination) are heuristics, without any convergence, nor performance guarantees.

In my thesis work, we observe that the DTN environment confers common characteristics to networking problems: (i) globally optimal solutions (e.g., paths, set of relays) are sought for, but each node acts independently at a local level; (ii) global information about the current solution is highly desirable in order to guarantee convergence, yet nodes only have local information. These characteristics form a generic problem (underlying most DTN networking problems) with two foci: (i) distributed optimization and (ii) distributed estimation. We believe that a solid understanding of the tradeoffs involved, and efficient solutions to both aspects of the generic problem, will reveal the advantages and limitations of various algorithms for specific Opportunistic Networking problems, and will allow for provably optimal solutions to be derived. It is therefore the goal of this thesis to better understand and solve both aspects of the generic problem.

The result will be algorithms for both distributed optimization and distributed estimation in the context of Opportunistic Networks. In addition, we will also develop a generic analytical model and framework for distributed optimization and one for distributed estimation, that will predict the performance of our algorithms, as well as that of previous schemes.