This project aims at contributing to the design and performance evaluation of new transport and routing protocols in wireless and mobile ad hoc networks. These issues are of paramount importance to allow the realistic implementations and deployment of such promising networks which are based on the so-called cooperative communications schemes. We will use network coding to design systems achieving the goals. Network coding is a new research area which entails a new conception of networks and has been shown to be beneficial to a lot of cost functions involved in wired and wireless network functioning.
First we will consider the problem of designing transport protocols, i.e., protocols that ensure reliable data transfer (some feedback must allow the source to know whether the packet has made it successfully to the destination) as well as congestion control (management of buffer occupation at intermediate nodes). Transport protocols are crucial in networks for, e.g., file sharing or media distribution. TCP/IP is by far the most deployed Internet protocol for data transfer. As a first step toward wireless ad hoc mobile networks, we will address the problem of enhancing the behavior of the well-known TCP protocol in infrastructure networks with wired and wireless sections (WLAN based on the 802.11 standard). Indeed, TCP is known to perform poorly in wireless environments. Our goal is to design a network coding-based transport protocol that will benefit bandwidth offered by the physical layer by well-behaving over as more as possible wireless channels.
Going one step further, in the second task, we will address the same problem of reliable data transfer in DT-MANETs. Owing to the very random trajectory of each packet from the source to the destination, TCP/IP turns out to be very inefficient for reliable data transfer over DT-MANETs. In addition, acknowledging each arriving packet waists much resource in terms of energy and memory. We will propose new efficient protocols substituting TCP/IP over DT-MANETs. In particular, we will look at the multicast case (one or several sources to one or several destinations).
In the third task, the goal is to design optimal routing policies for transmission in DTN, in
terms of different objectives: delay, throughput, energy consumption and memory. Reliable data transfer in DT-MANETs is a difficult and important problem as it controls buffer nodes’ occupation, and hence, owing to the limited nodes resource in memory, determines network well-functioning We will focus on policy optimization in order to save energy consumed at mobile terminals, while preserving delay performance. We will first assume random mobility models, and then more and more refined mobility models so as to account for realistic clustered mobility models of heterogeneous nodes, i.e., nodes with different networking abilities that have a limited number of home points and do not scatter uniformly over the space, as assumed with random mobility models.
The fourth purpose of this project is to find bounds on the distribution of MANETs capacity.
This tasks aims at providing confidence intervals on the capacity of MANETs, which is of paramount importance to predict performance. In order to be more accurate than exiting scaling laws for capacity of MANETs that are valid only for infinite time and when the size of the network tends to infinity, we will target our efforts towards bounding the probability that a given capacity will be achieved in a limited time period, given the mobility model and the size of the network. In the same way as in the previous task, we will consider more and more refined mobility models so as to foresee what capacity,
defined in terms of a specific graph min-cut that will be defined, can be expected in terms of the transmission duration.
It is worth noting that in Tasks 2 to 4 aims at solving transport, routing and capacity determination in DT-MANETs where nodes cannot assume any knowledge of mobility patterns. In that case, accounting for realistic mobility models is of paramount importance to derive useful results and devise efficient protocols.