Sharif Digital Repository

The minimum cost subgraph for joint distributed source and network coding has been studied using a linear programming approach. This problem considers some statistically correlated information sources in a network, where sending over each link has a defined cost. The desired solution to this problem is a set of coding rates over the graph links, that minimizes the total communication costs. But in this problem, the placement of the sources is pre-defined, which is not necessarily optimal. In this paper, for the first time, we find a "good" placement for the sources using a genetic programming approach. This problem can be seen as the generalized mirror sites problem, where sources are not... 

A function is said to be additive if, similar to mutual information, expands by a factor of n, when evaluated on n i.i.d. repetitions of a source or channel. On the other hand, a function is said to satisfy the tensorization property if it remains unchanged when evaluated on i.i.d. repetitions. Additive rate regions are of fundamental importance in network information theory, serving as capacity regions or upper bounds thereof. Tensorizing measures of correlation have also found applications in distributed source and channel coding problems as well as the distribution simulation problem. Prior to our work only two measures of correlation, namely the hypercontractivity ribbon and maximal... 

In this paper we study the sum modulo two problem proposed by Körner and Marton. In this source coding problem, two transmitters who observe binary sources X and Y, send messages of limited rate to a receiver whose goal is to compute the sum modulo of X and Y. This problem has been solved for the two special cases of independent and symmetric sources. In both of these cases, the rate pair (H(X|Y), H(Y|X)) is achievable. The best known outer bound for this problem is a conventional cut-set bound, and the best known inner bound is derived by Ahlswede and Han using a combination of Slepian-Wolf and Körner-Marton's coding schemes. In this paper, we propose a new outer bound which is strictly... 

New inner and outer bounds on the achievable compression-equivocation rate region for generalised secure data compression with side information are given that do not match in general. In this setup, two senders, Alice and Charlie intend to transmit information to Bob via channels with limited capacity so that he can reliably reconstruct their observations. The eavesdropper, Eve, has access to one of the channels at each instant and is interested in the source of the same channel at the time. Bob and Eve also have their own observations, which are correlated with Alice's and Charlie's observations. In this model, two equivocation and compression rates are defined with respect to the sources... 

Multimedia multicast with two servers based on the mul- titerminal source coding is studied in some previous researches. Due to the possibility of providing an approach for practical code design for more than two correlated sources in IMTSC/CEO setup, in this paper, the framework of Slepian-Wolf coded quantization is extended and a practical code design is presented for IMTSC/CEO with the number of encoders greater than two. Then the multicast system based on the IMTSC/CEO is applied to the cases with three, four and five servers. Since the underlying code design approach for the IMTSC/CEO problem has the capability of applying to an arbitrary number of active encoders, the proposed MMBMSC... 

We study the estimation of a Gaussian source by a Gaussian wireless sensor network (WSN) where L distributed sensors transmit noisy observations of the source through a fading Gaussian multiple access channel to a fusion center. In a recent work Gastpar, [ "Uncoded transmission is exactly optimal for a Simple Gaussian Sensor Network", IEEE Trans. Inf. Theory, vol. 54, no. 11, pp. 52475251, Nov. 2008] showed that for a symmetric Gaussian WSN with no fading, uncoded (analog) transmission achieves the optimal performance theoretically attainable (OPTA). In this correspondence, we consider an asymmetric fading WSN in which the sensors have differing noise and transmission powers. We first... 

To perform software maintenance, developers must have a relatively good understanding of the program's source code, which is often written by other developers. Code summaries, which describe a program's entities (for example, its methods), help developers better comprehend code more quickly. However, generating code summaries can be challenging. To mitigate this problem, CrowdSummarizer exploits crowdsourcing, gamification, and natural-language processing to automatically generate high-level summaries of Java program methods. Researchers have implemented it as an Eclipse plug-in together with a Web-based code summarization game that can be played by the crowd. Two empirical studies... 

Watermarking is the process of hiding information in a media and usually used to for copyright protection. In software watermarking, the information is added to the source code of program and can be extracted from compiled executable code. In this paper, a new software watermarking algorithm is proposed by equation reordering. In this method, the order of operands in an equation is changed according to the information we want to hide in a way that the result of the equation is unchanged. One of the main advantages of this method is that no code added to the program, so the size and speed of the program not changed  

In this paper, we develop a new error-resistant secure lossless source coding scheme for discrete sources. Recent results indicate that the same turbo principle that provides sub-optimal strategy for channel coding, can be used to obtain efficient source coding schemes. We extend the source turbo-coding idea to include security for transmission of the compressed data and also to ensure the lossless recovery of information at the receiver. Compression and security are achieved by adapting the random puncturing strategy to the statistics of the source. Lossless compression is guaranteed by finely puncturing the encoded data of a parallel-concatenated turbo code while verifying the integrity of... 

The aim of this paper is introducing coding scheme that can potentially achieve the sum rate-distortion bound of Gaussian IMTSC/CEO problem with arbitrary number of encoders. No previous work has been reported on code design for the IMTSC/CEO problem with the number of encoders greater than two. In this paper the framework of Slepian-Wolf coded quantization is extended to the case with the number of encoders greater than two. On the other hand, we apply the approach to the case of unequal levels of the observation noise at the encoders. Encoding process consists of trellis coded quantization (TCQ) and LDPC codes. The proposed framework can be easily used for any number of active encoders... 

In this paper, we investigate reliable transmission of three correlated discrete memoryless sources over a two-relay network. In our considered model, one of the sources is available at the sender whereas, the other two sources are known to the first and the second relay. We present both joint and separate source-channel coding schemes, and derive the corresponding sets of sufficient conditions for reliable sources transmission. The manner of cooperation in both schemes is Decode-and-Forward strategy. In the joint approach, we generalize the correlation preserving mapping technique to our model using nested backward decoding. Our proposed separate approach is based on Slepian-Wolf source... 

In this work, we study the lossy source coding of a Gaussian source with polar codes. We show through two distributions on discrete alphabet, quantized approach as well as Central Limit Theorem (CLT) approach, when the alphabet size grows to infinity, polar codes can achieve the rate distortion bound for a Gaussian source. By comparing the rate of convergence on two distributions, we show that the quantized approach have a better convergence rate than the CLT approach  

Visual attributes are great means of describing images or scenes, in a way both humans and computers understand. In order to establish a correspondence between images and to be able to compare the strength of each property between images, relative attributes were introduced. However, since their introduction, hand-crafted and engineered features were used to learn increasingly complex models for the problem of relative attributes. This limits the applicability of those methods for more realistic cases. We introduce a deep neural network architecture for the task of relative attribute prediction. A convolutional neural network (ConvNet) is adopted to learn the features by including an... 

A matrix pseudo inversion based method is developed for coding of images in real field in a way that at the receiver, one can restore the lost portions of the images. The proposed solution is similar to channel coding techniques, but it is done before source coding at the transmitter and after decoding of the compressed data at the receiver. Experiments show the effectiveness of the proposed solution in recove ring missed portions of the images and video frames. ©2007 IEEE