智能电表集中抄表系统入户子系统 第2页
intelligence (AI) can be applied to augment the skills of a network planner in the area of greenfield access network planning.
The process of designing and planning new network implementations can be considered the first stage of the network management life cycle. Network planning is a skilled, labour-intensive process in which many different aspects of the network performance must be considered. For example, the basic functional performance of the network must be within design limits (e.g. transmission loss tolerances) and the link bandwidths must be dimensioned to handle forecast traffic and signalling overheads. The overall network availability must be calculated, based on estimated component reliabilities, and redundancy must be introduced if required. The network must also be designed for minimum cost, which may mean a relatively simple calculation of installed cost if a single technology is considered, but could mean comparisons of whole-life cost if technology comparisons are required. Traditionally, these various aspects are considered to a great extent separately and sometimes even sequentially, in an attempt to make manageable the complexities involved. Thus there has evolved many different process steps, planning rules, and species of network planners, investment appraisers, strategic planners, tactical and detail planners, copper, radio and fibre network specialists, traffic modellers, and reliability engineers. With all these planning steps and considerations, even experienced human experts have little chance of efficiently designing truly optimal networks. It is clear that, against the background of increasing demands on the performance of the networks and of the planners, network planning is a discipline full of opportunities.
3. Optimising ‘total’ performance
The way forward is to exploit those very technologies that are giving rise to the problems of increased network complexity and customer demands. Computing power increasingly offers the opportunity to break away from the arbitrary fragmentations of traditional network planning, towards a more comprehensive and holistic approach. A reasonable definition of the domain of ‘total’ network performance that must be navigated is the functionality/reliability/minimum cost triangle. On this ‘map’ [2], whole-life cost is simply the overlap of cost and reliability, and optimum bandwidth (capacity) dimensioning can be considered to be a trade-off between functionality and cost. The three fundamental aspects of network performance — function, reliability, cost — can now feasibly be considered simultaneously and interactively, vastly improving the possibility of obtaining optimal design solutions, even for complex networks. Unified network models, capturing the ‘total’ performance characteristics of the network, could become a powerful tool, not just for network planners and managers, but also operational and maintenance engineers, sales people, and marketers.
If this vision of more unified and universally useful network models is to be realised, then of course more than just the performance of the network ‘machine’ must be considered. Extra types of data, such as customer socio-economics, and billing or revenue data, could be required to segment and address the target markets. Also, the effect of human factors on network performance can be important, for example, most telecommunications companies are aware of the deleterious effect of human intrusion in the access network. Increasingly even these complexities can be, and must be, considered in planning the competitive networks of the future .
4. Systems developments and trends
How then are we to best exploit the increasing power of our computing hardware and software, to move towards comprehensive network planning and ‘total’ performance? Several trends have clearly become important in the last few years. Object-oriented programming and data structures with their inherent characteristics of encapsulation, inheritance and polymorphism are ideally suited to the efficient modelling of telecommunications networks . ‘Layered’ network data models which describe the hierarchy and containment of network objects (e.g. duct-cable-bearer-channel) are crucial to any powerful planning model. A commonly shared, readily accessible cache of network plant and performance data is essential if wasteful data duplication is to be avoided. User-friendly interfaces are as always required, and this usually means at least a graphical user interface (GUI) and often needs a geographical information system (GIS) approach. A general systems architecture which shares data, but allows a diversity of client applications and presentation layers, offers the best compromise between data integrity and application quality . Rapid application development techniques are increasingly powerful, and the resulting applications, if modular and coupled with ‘core’ data, can produce powerful results. The requirements can be summarised as a set of network performance models and planning tools which are:
• modular,
• distributed,
• interconnected,
• based on shared data,
• easy to access and use,
• holistic and comprehensive.
‘Thick’ client systems, sharing data via intranets, are increasingly likely to meet these requirements. Expert system applications are being developed which can solve difficult network optimisation problems in a few seconds, running on a desktop PC .
5. Examples of developmental tools and systems
Several examples of such recent developments, in use within BT, illustrate the possibilities. In varying degrees, they are expert systems — they embody the skills and rules employed by skilled human planners, and their impact on planning productivity and quality is proving to be extremely significant.
5.1 BECS
The first, BECS (Budgetary Estimate Calculation System), is a medium-size (30 000 line) Visual Basic (VB) application which allows planners to estimate
上一页 [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] ... 下一页 >>
智能电表集中抄表系统入户子系统 第2页下载如图片无法显示或论文不完整，请联系qq752018766
