We present a distributed random linear network coding approach for transmission and compression of information in general multisource multicast networks. Network nodes independently randomly select mappings from inputs onto output links over some field. show that this achieves capacity with probability exponentially approaching 1 the code length. also demonstrate performs when necessary network, generalizing error exponents Slepian-Wolf natural way. Benefits are decentralized operation...

A novel randomized network coding approach for robust, distributed transmission and compression of information in networks is presented, its advantages over routing-based approaches demonstrated.

We consider the problem of establishing minimum-cost multicast connections over coded packet networks, i.e., networks where contents outgoing packets are arbitrary, causal functions received packets. both wireline and wireless as well static (where membership group remains constant for duration connection) dynamic changes in time, with nodes joining leaving group). For multicast, we reduce to a polynomial-time solvable optimization problem, present decentralized algorithms solving it. These...

Network coding substantially increases network throughput. But since it involves mixing of information inside the network, a single corrupted packet generated by malicious node can end up contaminating all reaching destination, preventing decoding. This paper introduces <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">the</i> xmlns:xlink="http://www.w3.org/1999/xlink">first</i> xmlns:xlink="http://www.w3.org/1999/xlink">distributed</i>...

Distributed randomized network coding, a robust approach to multicasting in distributed settings, can be extended provide Byzantine modification detection without the use of cryptographic functions is presented this paper.

We consider energy efficient network coding design in wireless networks with multiple unicast sessions. Our approach decomposes sessions into a superposition of multicast and sessions, occurring only within each session. give an optimization that is more general than the existing poison-remedy formulation. For case wireless, we XOR achievable rate region for primary interference model. To simplify operation, oblivious backpressure algorithm which does not optimize overhearing transmissions,...

We design codes to transmit information over a network, some subset of which is controlled by malicious adversary. The computationally unbounded, hidden adversary knows the message be transmitted, and can observe change part network being controlled. nodes do not share resources such as shared randomness or private key. first consider unicast problem in with |epsiv parallel, unit-capacity, directed edges. rate-region has two parts. If controls fraction p < 0.5 edges, maximal throughput...

Precise connections between algebraic network coding and flows are explained in this article. Our combinatorial formulations offer new insights, mathematical simplicity, lead to a substantially tighter upper bound on the field size required for given connection problem than that R. Koetter et al.

We present an information-theoretic framework for network management recovery from nonergodic link failures. Building on recent work in the field of coding, we describe input-output relations nodes terms codes. This very general concept behavior as a code provides way to quantify essential information that needed switch among different codes (behaviors) failure scenarios. compare two types schemes, receiver-based and network-wide, consider formulations quantifying management. The first is...

In this paper we show how information contained in robust network codes can be used for passive inference of possible locations link failures or losses a network. For distributed randomized coding, bound the probability being able to distinguish among given set failure events, and give some experimental results one two randomly generated networks. We also required field size complexity designing code that distinguishes events

We consider the problem of optimally allocating a given total storage budget in distributed system. A source has data object which it can code and store over set nodes; is allowed to any amount each node, subject constraint. collector subsequently attempts recover original by accessing random fixed-size subset these nodes. Successful recovery occurs when coded this nodes at least size object. The goal determine node so that probability successful maximized. solve high regime. Our results be...

We investigate the problem of using several storage nodes to store a data object, subject an aggregate budget or redundancy constraint. It is challenging find optimal allocation that maximizes probability successful recovery by collector because large space possible symmetric and nonsymmetric allocations, nonconvexity problem. For special case probability-1 recovery, we show minimizes required symmetric. further explore access models, determine in high-probability regime for interest. Based...

Enabled by the availability of both excitatory and inhibitory post-synaptic currents, a biological neural network is inherently capable implementing more sophisticated non-monotonic classification schemes. While such currents are readily emulated in software-based artificial using positive negative synaptic weighting factors, same not as straight forward hardware implementation. In this work, two dual-gate thin-film transistors with effectively infinite direct-current input impedance...

We present an information theoretic framework for network management recovery from nonergodic link failures. Building on recent work in the field of coding, we describe input-output relations nodes terms codes. This very general concept behavior as a code provides fundamental way to quantify essential that needed switch among different codes (behaviors) failure scenarios. give bounds various connection problems, basic parameters such number source processes and links minimum source-receiver...

We show a statistical pruning approach for maximum likelihood (ML) detection of multiple-input multiple-output (MIMO) systems. present general strategy sphere decoder (SD), which can also be applied to any tree search algorithms. Our rules are effective especially the case when SD has high complexity. Three specific given and discussed. From analyzing union bound on symbol error probability, we that diversity order deterministic is only one by fixing probability. By choosing different...

A general description of the behavior a communication network is given by codes specifying relationships between each node's observable source processes and its input output signals. In this work we focus on acyclic multicast networks, consider two formulations for quantifying management link failure recovery, in terms selecting codes.