No description available.

Simulation has become an important way to observe and understand various networking phenomena under various conditions. As the demand to simulate larger and more complex networks increases, the limited computing capacity of a single workstation and the limited simulation capability of a single network simulator have become apparent obstacles to the simulationists. In this research we develop techniques that can scale a simulation to address the limited capacity of a single workstation, as well as techniques that can compose a simulation from different simulator components to address the limited capability of a single network simulator. We scale a simulation with two different approaches: 1) We reduce the resource requirement of a simulation substantially, so that larger simulations can fit into one single workstation. In this thesis, we develop three technqiues (Negative Forwarding Table, Multicast Routing Object Aggregation and NIx-Vector Unicast Routing) to aggregate and compress the large amount of superfluous or redundant routing state in large multicast simulations. 2) The other approach to scale network simulations is to partition a simulation model in a way that makes the best use of the resources of the available computer cluster, and distribute the simulation onto the different processors of the computer cluster to obtain the best parallel simulation performance. We develop a novel empirical methodology called BencHMAP (Benchmark-Based Hardware and Model Aware Partitioning) that runs small sets of benchmark simulations to derive the right formulas of calculating the weights that are used to partition the simulation on a given computer cluster. On the other hand, to address the problem of the limited capability of a network simulator, we develop techniques for building complex network simulations by composing from independent components. With different existing simulators good at different protocol layers/scenarios, we can make each simulator execute the layers where it excels, using a simulation backplane to be the interface between different simulators. In this thesis we demonstrate that these techniques enable us to not only scale up simulations by orders of magnitude with a good performance, but also compose complex simulations with high fidelity.

This work focuses on the performance characterization of distributed collaborative ad-hoc networks, focusing on such metrics as the lifetime, latency, and throughput capacity of two such classes of networks. The first part concerns modeling and optimization of static Wireless Sensor Networks, specifically dealing with the issues of energy efficiency, lifetime, and latency. We analyze and characterize these performance measures and discuss various fundamental design tradeoffs. For example, energy efficiency in wireless sensor networks can only be improved at the cost of the latency (the delay incurred during communication). It has been clearly shown that improvement in energy efficiency through data aggregation increases the latency in the network. In addition, sleep-active duty cycling of nodes (devices constituting the network), a commonly employed mechanism to conserve battery lifetime in such networks, has adverse effects on their functionality and capacity. Hence these issues deserve a detailed study. The second part of this work concerns performance modeling of Delay Tolerant Networks (DTNs) and Sparse Mobile Ad-Hoc Networks (SPMANETs) in general. We first investigate the effect of modern coding, such as the application of packet-level rateless codes, on the latency, reliability, and energy efficiency of the network. These codes provide us the means to break large messages into smaller packets thereby enabling efficient communication. The work then focuses on developing and formalizing an information-theoretic framework for Delay Tolerant- and other Sparse Mobile Networks. This is enabled by the use of an embedded-Markov-chain approach used for complex queuing-theoretic problems. An important goal of this work is to incorporate a wide range of mobility models into the analysis framework. Yet another important question will be the effect of changing the mobility on the comparative performance of networking protocols. Lastly, the framework will be extended to various communication paradigms such as two-hop vs multi-hop routing, unicast, and multicast.

This dissertation first studies two-dimensional wavelet codes (TDWCs). TDWCs are introduced as a solution to the problem of designing a 2-D code that has low decoding- complexity and has the maximum erasure-correcting property for rectangular burst erasures. The half-rate TDWCs of dimensions N₁ X N₂ satisfy the Reiger bound with equality for burst erasures of dimensions N₁ X N₂/2 and N₁/2 X N₂, where GCD(N₁,N₂) = 2. Examples of TDWC are provided that recover any rectangular burst erasure of area N₁N₂/2. These lattice-cyclic codes can recover burst erasures with a simple and efficient ML decoding. This work then studies the problem of distributed source coding for two and three correlated signals using channel codes. We propose to model the distributed source coding problem with a set of parallel channel that simplifies the distributed source coding to de- signing non-uniform channel codes. This design criterion improves the performance of the source coding considerably. LDPC codes are used for lossless and lossy distributed source coding, when the correlation parameter is known or unknown at the time of code design. We show that distributed source coding at the corner point using LDPC codes is simplified to non-uniform LDPC code and semi-random punctured LDPC codes for a system of two and three correlated sources, respectively. We also investigate distributed source coding at any arbitrary rate on the Slepian-Wolf rate region. This problem is simplified to designing a rate-compatible LDPC code that has unequal error protection property. This dissertation finally studies the distributed source coding problem for applications whose wireless channel is an erasure channel with unknown erasure probability. For these application, rateless codes are better candidates than LDPC codes. Non-uniform rateless codes and improved decoding algorithm are proposed for this purpose. We introduce a reliable, rate-optimal, and energy-efficient multicast algorithm that uses distributed source coding and rateless coding. The proposed multicast algorithm performs very close to network coding, while it has lower complexity and higher adaptability.