Search

We propose a LDPC Convolutional Code ensemble together with an expanding-window message-passing decoder that asymptotically have anytime properties when used for streaming transmission on the binary erasure channel. We show analytically that the decoding erasure probability of these codes decays exponentially over decoding delay and determine the corresponding anytime exponents., QC 20130118

The two-user discrete memoryless state-dependent multiple-access channel with cooperating encoders is considered. It is assumed that the channel is controlled with two independent states such that each of the channel state information is noncausally available at one encoder. Moreover, based on the intrinsic characteristic of wireless networks, it is assumed that each encoder strictly causally receive a noisy version of the transmitted signal of the other encoder. Hence, the encoders can cooperate in transmitting the message to the receiver. An achievable rate region for the channel is derived by use of rate splitting, block Markov encoding and Gelfand-Pinsker multilevel encoding along with partial decoding at the encoders and backward decoding at the receiver. Finally, for a Gaussian channel model the proposed achievable rate region is evaluated and discussed., QC 20130118

We propose a new unequal error protection (UEP) scheme for Luby Transform (LT) codes over additive white Gaussian noise (AWGN) channels. In contrast to the state-of-the-art UEP scheme for LT codes where the effective code rates are exploited for UEP, we characterize the variable-node degree distribution to achieve UEP for LT codes. For performance analysis, we compare the lower bounds on bit error rate for our proposed scheme and a conventional UEP-based LT codes over AWGN channels for various parameters. Moreover, it is demonstrated through numerical examples that our proposed UEP scheme has better performance than the conventional UEP-based LT codes over a wide range of code rates and channel conditions., QC 20130116

We propose Gaussian process dynamical models (GPDMs) as a new, nonparametric paradigm in acoustic models of speech. These use multidimensional, continuous state-spaces to overcome familiar issues with discrete-state, HMM-based speech models. The added dimensions allow the state to represent and describe more than just temporal structure as systematic differences in mean, rather than as mere correlations in a residual (which dynamic features or AR-HMMs do). Being based on Gaussian processes, the models avoid restrictive parametric or linearity assumptions on signal structure. We outline GPDM theory, and describe model setup and initialization schemes relevant to speech applications. Experiments demonstrate subjectively better quality of synthesized speech than from comparable HMMs. In addition, there is evidence for unsupervised discovery of salient speech structure., QC 20120308, LISTA

We propose Gaussian process dynamical models (GPDMs) as a new, nonparametric paradigm in acoustic models of speech. These use multidimensional, continuous state-spaces to overcome familiar issues with discrete-state, HMM-based speech models. The added dimensions allow the state to represent and describe more than just temporal structure as systematic differences in mean, rather than as mere correlations in a residual (which dynamic features or AR-HMMs do). Being based on Gaussian processes, the models avoid restrictive parametric or linearity assumptions on signal structure. We outline GPDM theory, and describe model setup and initialization schemes relevant to speech applications. Experiments demonstrate subjectively better quality of synthesized speech than from comparable HMMs. In addition, there is evidence for unsupervised discovery of salient speech structure., QC 20120308, LISTA

We propose Gaussian process dynamical models (GPDMs) as a new, nonparametric paradigm in acoustic models of speech. These use multidimensional, continuous state-spaces to overcome familiar issues with discrete-state, HMM-based speech models. The added dimensions allow the state to represent and describe more than just temporal structure as systematic differences in mean, rather than as mere correlations in a residual (which dynamic features or AR-HMMs do). Being based on Gaussian processes, the models avoid restrictive parametric or linearity assumptions on signal structure. We outline GPDM theory, and describe model setup and initialization schemes relevant to speech applications. Experiments demonstrate subjectively better quality of synthesized speech than from comparable HMMs. In addition, there is evidence for unsupervised discovery of salient speech structure., QC 20120308, LISTA

We propose Gaussian process dynamical models (GPDMs) as a new, nonparametric paradigm in acoustic models of speech. These use multidimensional, continuous state-spaces to overcome familiar issues with discrete-state, HMM-based speech models. The added dimensions allow the state to represent and describe more than just temporal structure as systematic differences in mean, rather than as mere correlations in a residual (which dynamic features or AR-HMMs do). Being based on Gaussian processes, the models avoid restrictive parametric or linearity assumptions on signal structure. We outline GPDM theory, and describe model setup and initialization schemes relevant to speech applications. Experiments demonstrate subjectively better quality of synthesized speech than from comparable HMMs. In addition, there is evidence for unsupervised discovery of salient speech structure., QC 20120308, LISTA

We investigate the error floor performance of Luby Transform (LT) codes over the additive white Gaussian noise channel. We first derive a lower bound on the bit error rate for an LT code, which we subsequently use to show that the corresponding error floor is predominantly caused by low-degree variable nodes. Based on this observation, we propose a modified encoding scheme for LT codes that provides a lower error floor with no increase in encoding and decoding complexities. The convergence behavior of the proposed scheme is analyzed using extrinsic information transfer charts, and shown to be similar to the original LT code. Numerical examples demonstrate the improvements of the modified LT code as a stand-alone code and as a component code of a Raptor code., QC 20120206

We consider a base-station broadcasting a set of order-insensitive packets to a user population over packet-erasure channels. To improve efficiency we propose a relay-aided transmission scheme using instantaneously-decodable binary network coding. Our proposed scheme ensures that a coded packet can be immediately decoded at the user side without delay. Moreover, only binary operations are required in the encoding and decoding processes, which decrease the computational complex. We further analyze the performance of the resulting broadcast scheme, and show that significant improvements in transmission efficiency are obtained as compared to previously proposed ARQ and network-coding-based schemes., QC 20120228

We study network coding for wireless networks with finite channel state information (CSI) at intermediate nodes (relays). Based on the CSI of the relay-sink channels, we adapt the network codes at the relays. For a specific network with two sources, four relays and two sinks, the analytic results show that one bit CSI of all relay-sink channels (global CSI) can reduce complexity (field size), and simultaneously decrease the erasure probability. Then, we generalize the results to relay networks with M users, N relays and J sinks. We show that fixed network codes without CSI cannot achieve instantaneous min-cut, i.e., min-cut under current channel state. We also show that with one bit global CSI, we can achieve instantaneous min-cut by adapting the network codes using an alphabet size L, where L is the number of sinks connecting to a relay. Yet, the fixed MDS network codes use an alphabet size L(M-1N-1). For the networks with perfect or imperfect source-relay channels, adaptive network codes with one bit global CSI have lower erasure probability than the codes without CSI. Thus, one bit global CSI can reduce the erasure probability, and simultaneously reduce coding complexity., QC 20120416

In distributed storage systems reliability is achieved through redundant storage nodes distributed in the network. Then a data collector can recover source information even if some nodes fail. To maintain reliability, an autonomous and efficient protocol should be used to reconstruct the failed node. Therepairprocess causes traffic in the network. Recent results in e.g., [1], [2] found the optimal traffic-storage tradeoff, and proposed regenerating codes to achieve the optimality. We investigate the link costs and the impact of network topologies during therepairprocess. We formulate the minimum costrepairproblem in joint and decoupled methods. We investigate the required field size for the joint method. For the decoupled method, we show that the optimization problem is linear for the linear cost. We further show that the cooperation of surviving nodes could efficiently exploit the network topology and reduce therepaircost. The numerical results in tandem, star and grid networks show the benefits of our methods in term of the repair cost., QC 20120208

A class of semi-deterministic relay channels is introduced and it is shown that for channels in the class capacity can be achieved using an instantaneous relaying scheme. Applications to Gaussian channels are then discussed., QC 20120203

In this work, we study the MIMO Gaussian bidirectional broadcast channel (BBC) with common message and characterize the capacity region. Moreover, we show that the transmit covariance matrix optimization problem has the same structure as the corresponding optimization problem of the BBC without common message which leads to the comfortable position to transfer results from one scenario to the other. This problem is motivated by the concept of bidirectional relaying in a three-node network, where a half-duplex relay node establishes a bidirectional communication between two other nodes and thereby adds an own multicast message to the communication., QC 20111219

We determine the rate region of a source coding problem with common reconstruction and action-dependent side information where an action sequence is taken by an encoder over a rate-limited link. We show that the rate region depends only on the sum-rate and the sum-rate distortion and cost function is characterized. The result serves as a fundamental limit in transmission scenarios where the encoder wants to control and monitor the quality of the decoder's reconstruction via the respective uses of action sequences and a common reconstruction constraint., QC 20111117

We design non-codebook-based Multiple-Input Multiple-Output (MIMO) precoding schemes using multiple cell-specific reference signals patterns for the time division duplex (TDD) mode of LTE, where channel reciprocity can be exploited. Previously proposed non-codebook-based precoding schemes typically use UE specific reference signals for demodulation. Cell specific reference signals are however always allocated for the transmission of common control signalling, mobility measurements and downlink channel quality measurements. In order to save the resources occupied by UE specific reference signals, and to simplify UE implementation, a precoder-estimation scheme and a codebook-assisted scheme are designed. The codebook-assisted scheme has the novelty of using a codebook in equalization. Link throughput simulations indicate that the codebook-assisted scheme is preferable compared to the precoder-estimation scheme and the scheme using UE specific reference signals in scenarios with moderate UE velocity., QC 20140901

We study the effect of power adaptation on the outage diversity of Incremental-redundancy automatic-repeat-request (INR-ARQ) transmission over the multi-input multi-output (MIMO) block-fading channel. In particular, we derive the optimal outage diversity achieved by power adaptation in INR-ARQ systems, and show that power adaptation provides significant gains in outage diversity. We also prove that the optimal outage diversity is achievable with random coding schemes., QC 20120216

We consider multi-constrained power, rate and channelallocation crafted for low power consumption, delaytolerant traffic, and under interfering link conditions thatmay be used in a cognitive radio system. Specifically, aniterative distributed algorithm, based on a sum-powerconstrained sum-rate maximization with upper (andlower) per user and channel power and rate constraints,as well as upper per user sum-power and sum-rateconstraints is developed. The feasibility and performanceof the algorithm is demonstrated by simulation in acellular system. Simulations show that the multipleconstraints are handled while improving the sum-rate vs. sum-power relative an “equal power adaptive rate” RRM approach., © 2009 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.Qc 20120203

We consider multi-constrained power, rate and channelallocation crafted for low power consumption, delaytolerant traffic, and under interfering link conditions thatmay be used in a cognitive radio system. Specifically, aniterative distributed algorithm, based on a sum-powerconstrained sum-rate maximization with upper (andlower) per user and channel power and rate constraints,as well as upper per user sum-power and sum-rateconstraints is developed. The feasibility and performanceof the algorithm is demonstrated by simulation in acellular system. Simulations show that the multipleconstraints are handled while improving the sum-rate vs. sum-power relative an “equal power adaptive rate” RRM approach., © 2009 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.Qc 20120203

We consider multi-constrained power, rate and channelallocation crafted for low power consumption, delaytolerant traffic, and under interfering link conditions thatmay be used in a cognitive radio system. Specifically, aniterative distributed algorithm, based on a sum-powerconstrained sum-rate maximization with upper (andlower) per user and channel power and rate constraints,as well as upper per user sum-power and sum-rateconstraints is developed. The feasibility and performanceof the algorithm is demonstrated by simulation in acellular system. Simulations show that the multipleconstraints are handled while improving the sum-rate vs. sum-power relative an “equal power adaptive rate” RRM approach., © 2009 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.Qc 20120203

We consider multi-constrained power, rate and channelallocation crafted for low power consumption, delaytolerant traffic, and under interfering link conditions thatmay be used in a cognitive radio system. Specifically, aniterative distributed algorithm, based on a sum-powerconstrained sum-rate maximization with upper (andlower) per user and channel power and rate constraints,as well as upper per user sum-power and sum-rateconstraints is developed. The feasibility and performanceof the algorithm is demonstrated by simulation in acellular system. Simulations show that the multipleconstraints are handled while improving the sum-rate vs. sum-power relative an “equal power adaptive rate” RRM approach., © 2009 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.Qc 20120203