Stephan Rein

Low-Memory Wavelet Transforms for Wireless Sensor Networks: A Tutorial

The computational and memory resources of wireless sensor nodes are typically very limited, as the employed low-energy microcontrollers provide only hardware support for 16 bit integer operations and have very limited random access memory (RAM). These limitations prevent the application of modern signal processing techniques to pre-process the collected sensor data for energy and bandwidth efficient transmission over sensor networks. This tutorial introduces communication and networking generalists without a background in wavelet signal processing to low-memory wavelet transform techniques. We first explain the one-dimensional wavelet transform (including the lifting scheme for in-place computation), the two-dimensional wavelet transform, as well as the evaluation of wavelet transforms with fixed-point arithmetic. Then, we explain the fractional wavelet filter technique which computes wavelet transforms with 16 bit integers and requires less than 1.5 kByte of RAM for a 256 × 256 gray scale image. We present case studies illustrating the use of these low-memory wavelet techniques in conjunction with image coding systems to achieve image compression competitive to the JPEG2000 standard on resource-constrained wireless sensor nodes. We make the C-code software for the techniques introduced in this tutorial freely available.

Voice quality evaluation in wireless packet communication systems: a tutorial and performance results for RHC

As wireless systems evolve toward supporting a wide array of services, including traditional voice service, using packet-switched transport, it becomes increasingly important to assess the impact of packet-switched transport protocols on voice quality, in this article we present a tutorial on voice quality evaluation for wireless packet-switched systems. We introduce an evaluation methodology that combines elementary objective voice quality metrics with a frame synchronization mechanism. The methodology allows networking researchers to conduct effective and accurate quality evaluation of packet voice. To illustrate the use of the described evaluation methodology and interpretation of the results, we conduct a case study of the impact of robust header compression (ROHC) on the voice quality achieved with real-time transmission of GSM encoded voice over a wireless link.

CATWOMAN: Implementation and Performance Evaluation of IEEE 802.11 Based Multi-Hop Networks Using Network Coding

This paper investigates the performance of network coding for an IEEE802.11 enabled meshed network. By means of basic setups the impact of the medium access control in combination with network coding is investigated. In contrast to prior work the network coding approach is tailored to commercial WiFi hardware without any special tweaks. The implementation of network coding is done on top of an existing routing scheme known as B.A.T.M.A.N. which has some inherent advantages to support network coding. We present schemes to utilize the B.A.T.M.A.N. routing to detect coding opportunities. One finding is that the performance gain for the well known Alice and Bob scenario using network coding is 60% compared to a pure relaying scheme. The software used in the presented measurement campaign is made publicly available.

RObust Header Compression (ROHC) Performance for Multimedia Transmission over 3G/4G Wireless Networks

RObust Header Compression (ROHC) has recently been proposed to reduce the large protocol header overhead when transmitting voice and other continuous media over IP based protocol stacks in wireless networks. In this paper we evaluate the real-time transmission of GSM encoded voice and H.26L encoded video with ROHC over a wireless link. For the voice transmission we examine the impact of ROHC on the consumed bandwidth, the voice quality, and the delay jitter in the voice signal. We find that for a wide range of error probabilities on the wireless link, ROHC roughly cuts the bandwidth required for the transmission of GSM encoded voice in half. In addition, ROHC improves the voice quality compared to transmissions without ROHC, especially for large bit error probabilities on the wireless link. The improvement reaches 0.26 on the 5-point Mean Opinion Score for a bit error probability of 10−3. For the video transmission we examine the impact of ROHC on the consumed bandwidth. We find that the bandwidth savings with ROHC depend on the quantization scale used for the video encoding and the video content and ranges between 5–40% for typical scenarios.

Performance evaluation of the fractional wavelet filter: A low-memory image wavelet transform for multimedia sensor networks

Existing image wavelet transform techniques exceed the computational and memory resources of low-complexity wireless sensor nodes. In order to enable multimedia wireless sensors to use image wavelet transforms techniques to pre-process collected image sensor data, we introduce the fractional wavelet filter. The fractional wavelet filter computes the wavelet transform of a 256x256 grayscale image using only 16-bit fixed-point arithmetic on a micro-controller with less than 1.5kbyte of RAM. We comprehensively evaluate the resource requirements (RAM, computational complexity, computing time) as well as image quality of the fractional wavelet filter. We find that the fractional wavelet transform computed with fixed-point arithmetic gives typically negligible degradations in image quality. We also find that combining the fractional wavelet filter with a customized wavelet-based image coding system achieves image compression competitive to the JPEG2000 standard.

Low-complexity compression of short messages

We describe a low-complexity scheme for lossless compression of short text messages. The method uses arithmetic coding and a specific statistical context model for prediction of single symbols. Our particular contribution is a simple yet effective approach for storing highly complex statistics in a succinct yet effective data model that can easily be trained by text data. The proposed model already gives good compression rates with a RAM memory size of 128 kByte, thus making lossless data compression with statistical context modeling readily applicable to small devices like wireless sensors or mobile phones

Comparison of analytical and measured performance results on network coding in IEEE 802.11 ad-hoc networks

Network coding is a promising technology that has been shown to improve throughput in wireless mesh networks. In this paper, we compare the analytical and experimental performance of COPE-style network coding in IEEE 802.11 ad-hoc networks. In the experiments, we use a lightweight scheme called CATWOMAN that can run on standard WiFi hardware. We present an analytical model to evaluate the performance of COPE in simple networks, and our results show the excellent predictive quality of this model. By closely examining the performance in two simple topologies, we observe that the coding gain results from the interaction between network coding and the MAC protocol, and the gap between the theoretical and practical gains is due to the different channel qualities of sending nodes. This understanding is helpful for design of larger mesh networks that use network coding.

Fractional Wavelet Filter for Camera Sensor Node with external Flash and extremely little RAM

This paper introduces the fractional wavelet filter as a technique to compute fractional values of each wavelet subband, thus allowing a low-cost camera sensor node with less than 2 kByte RAM to perform a multi-level 9/7 picture wavelet transform. The picture dimension can be 256x256 using fixed-point arithmetic and 128x128 using floating-point arithmetic. The technique is applied to a typical 16 Bit sensor node architecture with external flash memory, which allows to line-wisely read and write picture data.

Identifying the classical music composition of an unknown performance with wavelet dispersion vector and neural nets

As the internet search evolves toward multimedia content based search and information retrieval, audio content identification and retrieval will likely become one of the key components of next generation internet search machines. In this paper we consider the specific problem of identifying the classical music composition of an unknown performance of the composition. We develop and evaluate a wavelet based methodology for this problem. Our methodology combines a novel music information (audio content) descriptor, the wavelet dispersion vector, with neural net assessment of the similarity between unknown query vectors and known (example set) vectors. We define the wavelet dispersion vector as the histogram of the rank orders obtained by the wavelet coefficients of a given wavelet scale among all the coefficients (of all scales at a given time instant). We demonstrate that the wavelet dispersion vector precisely characterizes the audio content of a performance of a classical music composition while achieving good generalization across different performances of the composition. We examine the identification performance of a combination of 39 different wavelets and three different types of neural nets. We find that our wavelet dispersion vector calculated with a biorthogonal wavelet in conjunction with a probabilistic radial basis neural net trained by only three independent example performances correctly identifies approximately 78% of the unknown performances.

Wavelet Image Two-Line Coder for Wireless Sensor Node with Extremely Little RAM

This paper gives a novel wavelet image two-line (Wi2l) coder that is designed to fulfill the memory constraints of a typical wireless sensor node. The algorithm operates line-wisely on picture data stored on the sensors flash memory card while it requires approximatively 1.5 kByte RAM to compress a monochrome picture with the size of 256x256 Bytes. The achieved data compression rates are the same as with the set partitioning in hierarchical trees (Spiht) algorithm. The coder works recursively on two lines of a wavelet subband while intermediate data of these lines is stored to backward encode the wavelet trees. Thus it does not need any list but three small buffers with a fixed dimension. The compression performance is evaluated by a PC-implementation in C, while time measurements are conducted on a typical wireless sensor node using a modified version of the PC-code.