Info

(Figure 10.8). Representative illustrations of ray diagrams and TL have already been presented elsewhere in this book. These examples are presented for illustrative purposes only and are not intended to portray a complete model run.

The reverberation versus time plots illustrated in Figure 10.6 are valid for a deep-water environment with a shallow sonic layer depth, a critical depth and sufficient depth excess to support convergence zone propagation, as well as surface duct and bottom bounce paths. At about 11 s, a strong bottom bounce return is evident in the bottom reverberation curve. At about 63 s, a strong CZ return is evident in both the volume and surface reverberation curves (but not in the bottom reverberation curve).

The SNR plot in Figure 10.7 indicates a high (positive) ratio at ranges corresponding to direct path (0-2 km) and convergence zone (near 50 km) detections. These sonar detection opportunities are vividly demonstrated in Figure 10.8. Here, high probabilities of detection are associated with direct-path and CZ ranges.

It is sometimes useful to plot different combinations of active sonar parameters on the same graph. From Table 10.2, for example, echo signal level can be plotted together with the noise and reverberation masking levels, as in Figure 10.9. With such information, signal excess as a function of range can be computed, as in Figure 10.10, which provides a direct indication of sonar detection performance. Specifically, all areas under the signal excess curve, but above the horizontal zero signal excess line, are associated

Range (nm)

Figure 10.9 Sample graphical combination of selected sonar parameters. See Table 10.2 for an explanation of terms.

Range (nm)

Figure 10.9 Sample graphical combination of selected sonar parameters. See Table 10.2 for an explanation of terms.

Range (nm)

Figure 10.10 Signal excess as a function of range based on the data presented in Figure 10.9. Active sonar detection ranges are indicated by regions of positive signal excess.

Range (nm)

Figure 10.10 Signal excess as a function of range based on the data presented in Figure 10.9. Active sonar detection ranges are indicated by regions of positive signal excess.

with detection opportunities. That is, detection is accomplished when the signal excess is greater than, or equal to, zero. Thus, NISSM and similar models can provide graphic assessments of active sonar performance under the stated environmental and system-specific conditions. The intermediate quantities provide diagnostic tools for analyzing particular aspects of active sonar design and operation problems.

10.4 Model operating systems

The trend toward modular designs in the programming of underwater acoustic models is evidenced by continued interest in what are here termed model operating systems (MOS) (e.g. Eller, 1990; Holmes et al, 1990). These systems provide a framework for the direct linkage of datamanagement software with computer-implemented codes of acoustic models. Such systems relieve model operators of much of the tedium associated with data entry operations.

Model operating systems facilitate comparative model evaluations (see Chapter 11) by standardizing the hardware and software configurations of different modeling techniques. The resulting uniformity also encourages a higher degree of configuration management and thus assists in the process of model evaluation. Sonar trainers now commonly utilize sophisticated range-dependent acoustic models to generate realistic training environments for the operators of sonar equipment (Miller, 1982,1983). Requirements for near-real-time program execution are frequently satisfied by using modular approaches similar to those employed in the construction of MOS.

In general, the development of sonar model operating systems is based on several considerations (Locklin and Webster, 1980): (1) it is assumed that the complex sonar performance modeling problem can be decomposed into a set of generic functions, each of which solves a well-defined and bounded portion of the overall problem (decomposition is a powerful technique to break down large problems into manageable units); (2) it is assumed that the problem can be solved on a computer through the use of an automated data system and that the development of this system can be accomplished such that its computational structure bears a precise relationship to the problem's decomposed functions; and (3) it is assumed that the functions implemented on a computer have the means to communicate with each other in a prescribed manner.

10.4.1 System architecture

Model operating systems fall into two general classes of software architecture: bundled and executive. The principal distinction between these two architectures concerns the packaging of the software and databases to meet specific modeling requirements. Bundled systems are tailored to a narrow range of modeling applications and, in general, have only one method implemented to perform a generic modeling function. Executive systems are tailored to respond to a broader range of modeling applications by allowing the user to interactively control the hierarchical selection of functions. Furthermore, executive systems allow for several methods of performing a single generic function, thus providing a dynamic formulation of a modeling application.

Two different modeling environments can be addressed by the MOS: research and production. These two environments are not different with respect to what computing functions are performed, but rather how the analyst uses the computing functions. This difference in usage should be reflected in the MOS architecture in order to provide the analyst with an optimized capability to perform both types of activities, if necessary. These environments are briefly discussed below to provide potential users with a perspective and a choice of the appropriate style of computing suitable to their specific needs.

The research environment entails activities such as experimentation, iterative problem solutions, computation of intermediate model results and model input sensitivity studies, among others. Practical applications include sonar system design and environmental-acoustics research. This implies that the MOS must be highly interactive with respect to the interface between the analyst and the MOS, and usually suggests an executive architecture.

The production environment is typified by operations that are definable, structurally organized and ordered. To a large extent, these activities, once defined, have specific end products, solve a complete problem and are performed repetitively using the same (or slightly modified) functions. Practical applications include routine sonar performance predictions and sonar operational trainers. This implies that the computing tools can be organized and structured beforehand, and then used repetitively thereafter. It further indicates that a more optimal composition of functions may be needed, with less user interactive capability and less flexibility with respect to the computing function composition. This usually suggests a bundled architecture.

The functional organization of the MOS features a uniform way of viewing the system structure for solution of both the passive and active sonar equations. While these two problems are distinctly different, this need not imply that the active MOS and passive MOS must be constructed as two independent systems. Rather, the redundancy in system control software and in certain lower level functions common to both the active and passive sonar problems makes separate approaches less attractive. For illustrative purposes, a hypothetical functional decomposition of both the passive and active sonar equations will be described below.

10.4.2 Sonar modeling functions

A representative decomposition of the sonar equations into generic modeling functions is shown in the system data flows in Figure 10.11 for the passive sonar equation (Locklin and Webster, 1980) and in Figure 10.12 for the active sonar equation (Locklin and Etter, 1988). These particular decompositions are appropriate for incorporation into an executive architecture.

Although no decisionary logic options or looping controls are shown in these figures, a prerequisite processing sequence is nonetheless established. This sequence starts with an extraction process operating on databases that results in the generation of data sets specific to the analyst's application. In the case of passive sonars (Figure 10.11), subsequent calculations use the extracted data to develop the data sets required by either the beam noise (BN) or beam signal (BS) functions, which contribute to the final sonar performance (SP) function. This particular conceptual picture identifies three methods for generating BN, and two methods for generating BS, although a greater or lesser number of methods could be accommodated. This reflects the various levels of complexity possible in noise modeling and passive sonar system parameterization, respectively.

( Target 77System 7/User specified f f Noise 7{ User 7( Ocean 7( Platform 7

I characteris- | characteris- i f beam pattern M sources ( | transmission i environmental | characteris- i tics tics data loss data data tics y (sl^ y y (sc) yy (BP) W (N) yy (tl) yy (oe) VV (sn V

Extract signal level/spectra (SL)

Was this article helpful?

0 0

Post a comment