Improvement of Signal Compression-Restoration Algorithms for Remote Measurement Systems

Abstract

612.317 N. V. Myasnikova, and B. V. Tsypin Real-time signal compression-recovery based on the Prony method is examined, and the complexity and time consuming nature of the calculations are demonstrated. An approach to the compression and recovery of signals based on a combination of the Prony method and an expansion in sign-varying components is described.The characteristic feature of modern remote measurement systems is a multichannel system for data acquisition followed by simultaneous transfer of a large number of measured values along a single communication channel. These information transmission systems have multiple channels. The number of channels in a system is determined by the number of independent data inputs. Figure 1 is a block diagram of a multichannel measurement system.The parameters P 1 (t), ..., P N (t) monitored by detectors D are converted into digital signals S 1 , ..., S N . A data acquisition and processing unit (DAS) forms packages ρ 1 , ..., ρ N of data in accordance with the output interface. The data packages ρ j can consist of compressed signals S j , j = 1, ..., N, which can be used to recover digital signals with the required accuracy.Compression is a necessary procedure in data acquisition systems for transmission along communication channels that can reduce the demands on the transmission channels or the cost of data transfer.A block diagram of a generalized algorithm for processing data in a DAS for the example of compressing an image along a single channel is shown in Fig. 2. As the basis, we take a certain rule for transfer of data along the channel, e.g., sampling with period T t or transfer of the data when nonstandard situations arise. The processing time T p must be shorter than the transfer time for the data, i.e., T p ≤ T t . In choosing the duration of data from the compressed frame, T c , it is necessary to take the estimated low-frequency data components of the signal into account: when the frame duration is longer, it is possible to determine the low-frequency components, but then the compression time is longer, T c = T t corresponds to real time operation.Very often, the compression problem is solved on the basis of an analytical description of the signals (processes), i.e., an approximation. Much attention has been devoted to the problems of approximating measurement signals [1]. These are the major approximation methods: polynomial; representing a signal by a system of linearly independent oscillations or by a system of orthonormal functions; expanding signals with a limited spectrum in a Kotelnikov series; and expanding signals in series of special functions (Legendre, Bessel, Haar, Chebyshev, etc.). The choice of the form of an analytic representation begins with a solution of one of the possible problems such as compression of signals, approximation by a small number of terms in a series (for a successful choice of the type of functions), a convenient form for spectral estimates, etc.In the technical literature, little attention is devoted to models for describin...