Abstract. A new one-pass algorithm for constructing dynamic Huffman codes is introduced and analyzed. We also analyze the one-pass algorithm due to Failer, Gallager, and Knuth. In each algorithm, both the sender and the receiver maintain equivalent dynamically varying Huffman trees, and the coding is done in real time. We show that the number of bits used by the new algorithm to encode a message containing t letters is < t bits more than that used by the conventional two-pass Huffman scheme, independent of the alphabet size. This is best possible in the worst case, for any one-pass Huffman method. Tight upper and lower bounds are derived. Empirical tests show that the encodings produced by the new algorithm are shorter than those of the other one-pass algorithm and, except for long messages, are shorter than those of the two-pass method. The new algorithm is well suited for on-line encoding/decoding in data networks and for file compression.

Memory is one of the most restrictedresources in many modern embedded systems. Code compression can provide substantial savings in terms of size. In a compressedcode CPU, acache miss triggers the decompression of a main memory block, before it gets transferred to the cache. Because the code must bedecompressible starting from any point #or at least at cache block boundaries#, most #le-orientedcompression techniques cannot be used. We propose two algorithms to compress code in a space-e#cient and simple to decompress way, one which is independent of the instruction set and another which depends on the instruction set. We perform experiments on two instruction sets, a typical RISC #MIPS# and a typical CISC #x86# and compare our results to existing #le-orientedcompression algorithms. 1 Introduction Manyembedded computing systems are space and cost sensitive. As a result, available memory is limited, posing serious constraints on program size. We are studying ways of reducing the size of...

We provide a tutorial on arithmetic coding, showing how it provides nearly optimal data compression and how it can be matched with almost any probabilistic model. We indicate the main disadvantage of arithmetic coding, its slowness, and give the basis of a fast, space-efficient, approximate arithmetic coder with only minimal loss of compression efficiency. Our coder is based on the replacement of arithmetic by table lookups coupled with a new deterministic probability estimation scheme.

This paper describes the architecture and implementation of a high-speed decompression engine for embedded processors. The engine is targeted to processors where embedded programs are stored incompressed form, and decompressed at runtime during instruction cache re ll. The decompression engine uses a unique asynchronous variable decompression rate architecture to process Huffman-encoded instructions. The resulting circuit is significantly smaller than comparable synchronous decoders, yet has a higher throughput rate than almost all existing designs. The 0.8 layout is all full-custom and contains predominantly dynamic domino logic. The top-level control, as well as several small state machines, are implemented using asynchronous logic. The design operates without a user-supplied clock. Simulations using Lsim show average throughput of 32 bits/45 ns on the output side, corresponding to about 480 Mbit/sec on the input side. The chip has been manufactured by MOSIS; tests show that the asynchronous implementation operates correctly, with an average throughput exceeding simulations: 32 bits/39 ns on the output side, corresponding to about 560 Mbit/sec on the input side. This speed isacceptable for our application. The area of the design (excluding the pad-frame overhead) is only 0.75 mm². The design is the first fabricated chip for an instruction decompression unit for embedded processors.

We consider the application of high volume information dissemination in broadcast based mobile environments. Since current mobile units accessing broadcast information have limited battery capacity, the problem of quick and energy-efficient access to data becomes particularly relevant as the number and sizes of information units increases. We propose several randomized and Huffman-encoding based indexing schemes that are sensitive to data popularity patterns to structure data transmission on the wireless medium, so that the average energy consumption of mobile units is minimized while trying to access desired data. We then propose an algorithm for PCS units to tune into desired data independent of the actual transmission scheme being used. We also empirically study the proposed schemes and propose different transmission modes for the base station to dynamically adapt to changes in the number of data files to be broadcasted, the available bandwidth and the accuracy of data popularity patterns.

. We present a Pascal implementation of the one-pass algorithm for constructing dynamic Huffman codes that is described and analyzed in a companion paper [Vitter, 1987]. The program runs in real time

Shannon's source-channel separation theorem holds only under asymptotic conditions, where both source and channel codes are allowed infinite length and complexity, which is not possible in practice. This observation has led to the increasing popularity of joint source-channel encoding and decoding schemes as viable alternatives for achieving reliable communication of signals across noisy channels. Joint source-channel encoders (JSCE) aim at designing the source and channel encoder of a system in some joint sense, while joint source-channel decoders (JSCD) concentrate on the design of the decoder. JSCD schemes have been

This paper investigates systems that steganographically embed information in the “noise ” created by automatic translation of natural language documents. The main thrust of the work focuses on two problems- generation of plausible steganographic texts, and avoiding transmission of the original source for stego objects. Because the inherent redundancy of natural language creates plenty of room for variation in translation, machine translation is ideal for steganographic applications. We describe the design and implementation of a scheme for hiding information in translated natural language text and present experimental results using the implemented system. While the initial work in this vein required the presence of both the source and the translation, the system detailed in this paper requires only the translated text for recovering the hidden message, increasing security and improving resource usage. These improvements occur not only because the source text is no longer available to the adversary, but also because a broader repertoire of defenses (such as mixing human and machine translation) can now be used. ∗ Preliminary and shorter versions of this work appeared in [21, 38]. 1 1

We show that high-resolution images can be encoded and decoded efficiently in parallel. We present an algorithm based on the hierarchical MLP method, used either with Huffman coding or with a new variant of arithmetic coding called quasi-arithmetic coding. The coding step can be parallelized, even though the codes for different pixels are of different lengths; parallelization of the prediction and error modeling components is straightforward.

The recent explosion in research on probabilistic data mining algorithms such as Bayesian networks has been focussed primarily on their use in diagnostics, prediction and efficient inference. In this paper, we examine the use of Bayesian networks for a different purpose: lossless compression of large datasets. We present algorithms for automatically learning Bayesian networks and new structures called "Huffman networks" that model statistical relationships in the datasets, and algorithms for using these models to then compress the datasets. These algorithms often achieve significantly better compression ratios than achieved with common dictionary-based algorithms such those used by programs like ZIP. 1 Introduction It has long been understood that even when confronted with a ten-gigabyte file containing data to be statistically analyzed, the actual information-theoretic amount of information in the file might be much less, perhaps merely a few hundred megabytes. This insight is curren...