Abstract

Introduction

Wireless sensor network is a distributed, self-organizing network multiple web of devices with sensors (sensors) connected to each other through a radio channel. Sensory Data is data received from the sensors (transducers). Their main feature is that they often have a numerical form and change in a fixed range. At the moment, there is an active development of this technology, because number of users increases continuously: in industry, housing and utilities sector, households, transport, protection.

1. Relevance of the topic

The main limitations in the use of sensor networks are:

  • Autonomy;
  • Lifetime;
  • Cost;
  • Form Factor [1].

Autonomy and lifetime limits can be solved by technical means as well as software. One way to solve this problem is data compression.

2. The purpose and objectives of the research, the expected results

The aim of the study is to increase the efficiency of storage and transmission of information received from the sensors.

The main objectives of the study:

  • to consider the compression algorithms of numerical data and to choose the most effective compression of sensory data.
  • to examine the input parameters and their influence to the outcome of the chosen methods.
  • to develop an automated system that is based on a brief analysis of the input data and that will determine the best parameters of the algorithms to compress them.

3. Scientific innovation in the application of

At the moment there are quite a number of algorithms for data compression. But most of them atr not appliable in this area. Thus, there is a reson to develop an algorithm that uses only integer operations in a reasonable working time and providing data compression without significant distortion. At the current stage, the problem is insufficiently studied: there are studies showing the effectiveness of this approach, but there are just a few works involved the study of algorithms.

4. The general formulation of the problem

Let's consider the formal statement of the problem. We suppose there is the initial sequence like

, (1)

where — is some number. We must define the functional F, which converts the original sequence into a sequence of bytes , as well as functional , performing the inverse functiona , where also is the numerical sequence like

, (2)

and the maximum error of this transformation must not exceed a certain predetermined threshold E:

, (3)

As a functional F will act our compression algorithm. The main problems in the determination of the algorithm are:

  • Fragmentation of data.
  • The limited computational resources
  • The absence of a math coprocessor.

4.1 Basic compression algorithms for numerical data

Before selecting of an algorithm we need to analyze the existing compression methods of numerical data They can be divided into 3 groups:

  • dictionary compression
  • entropy coding;
  • various methods for preliminary data processing [2].

Let's consider each group in more details.

4.2. The compression algorithms.

To further study the algorithm was chosen 3:

  • bits packaging. The basic idea of bits packing consists in using fewer amount of bits to encode values ​​of sequence elements. A new bit length for the values is ​​calculated using the following formula:
    , (4)

    where max is the maximum value in the sequence, min is the minimum value in a sequence, dx is a step of elements changing.

    Scheme of the mechanism of bits packaging
    Fig.1 Scheme of the mechanism of bits packaging

    Lets assume that we need to encode the sequence of shorts , and it is known that the numbers are ranging from 100 to 200. Thus, instead of the standard 16 bits, it suffices to use only 7.

  • Integer Wavelet Transform 5/3 [5].
  • Ramer-Douglas-Pekker algorithm [6].
Demonstration of the Ramer-Douglas-Pekker algorithm,
Pic.2 Demonstration of the Ramer-Douglas-Pekker algorithm
The size of the animation: 8 Кb
Count of frames: 5
The frame display: 1,5 sec per frame
Count of recurrence cycles: 7

The problem with real numbers can be solved in the following way: all the real number are pre-converted to an integer type, and then they are subject to the above-described methods of treatment.

4.3. Parameters of sensory data

As it was noted above, the sensor network is composed of nodes, each of which may be attached several sensors. The main parameters are the amount of sensors, which the values are obtained from​​, and the number of these values. Thus, there are two modes of operation:

  • it is passed a set of values ​​(50-100) from several sensors (3-5). In this case it is advisable to compress the sequence of each probe individually.
  • it is transmitted a multiple values ​​(2-5) from the large number of sensors (10-20). In this case, approach that will form the sequence of one-time measurements of different sensors will give greater efficiency [7].

In addition to various methods of sequences forming the result of compression will also be affected by the the compression algorithm and its parameters, which may depend on such factors as the amount of raw data measurements, minimum and maximum values, ​​required accuracy.

The conclusions

The main objective is to develop an automated system that can advance the analysis of raw data and select the algorithm parameters that provide the maximum compression ratio of the permissible time. Based on the above it can be concluded that the use of compression algorithms can reduce the power consumption sensor nodes. It is shown that depending on the input data is necessary to use different parameters compression algorithms. Thus, the direction of future work is the development and implementation of the rules of selection algorithm and its parameters for different sensor data compression algorithms, and checking their reality.

References:

  1. Садков А.В. Беспроводные сенсорные сети. Курс лекций, 2007 [Электронный ресурс]. Режим доступа: http://www.sumkino.com/wsn/course/

  2. Смирнов М.А., Обзор применения методов безущербного сжатия данных в СУБД, 2003 [Электронный ресурс]. — Режим доступа: http://compression.ru/download/articles/db/smirnov_2003_database_compression_review.pdf

  3. Ватолин Д., Ратушняк А., Смирнов М., Юкин В. Методы сжатия данных. Устройство архиваторов, сжатие изображений и видео. — М.: ДИАЛОГ—МИФИ, 2002. — 384 с.

  4. Статистическое двоичное кодирование источника [Электронный ресурс]. — Режим доступа: http://edu.dvgups.ru/METDOC/GDTRAN/YAT/TELECOMM/TEOR_PERED_SIGN/METOD/HARAK_SV/Stroev_3.htm

  5. Michael D. Adams, Faouzi Kossentini Reversible Integer-to-Integer Wavelet Transforms for Image Compression: Performance Evaluation and Analysis, 1999 [Электронный ресурс]. — Режим доступа: http://www.ece.uvic.ca/~frodo/publications/phdthesis.pdf

  6. David Douglas & Thomas Peucker, Algorithms for the reduction of the number of points required to represent a digitized line or its caricature [Электронный ресурс]. — Режим доступа: http://utpjournals.metapress.com/content/fm576770u75u7727/?genre=article&id=doi%3a10.3138%2fFM57-6770-U75U-7727

  7. Е.Г. Краморенко, М.В. Привалов, "Понижение энергопотребления сенсорных сетей за счет предварительной обработки данных", Информационные управляющие системы и компьютерный мониторинг 2013/ Сборник материалов к IV Всеукраинской научно-технической конференции студентов, аспирантов и молодых ученых. — Донецк, ДонНТУ — 2013, с. 364 — 369.

Resume