官术网_书友最值得收藏!

Computational deployment patterns

The Industrial Internet encompasses many topologies of assets, and sensor and computational distributions. Traditional control systems typically disperse computational power close to the physical systems, in the network periphery, and are performed at the controller level. Traditional data centers perform centralized computations. The Industrial Internet requires a combination of both of these. Smart sensors and devices perform dispersed computations to enable localized control and decision making, while concentrated computation placed throughout the network supports higher-level decision making.

Network connectivity is critical for communication between the periphery and the concentrated capabilities. Concentrated computational capabilities enable a wider and deeper spectrum of analytics using data integrated throughout the industrial system, including traditional applications data.

The placement of computational capabilities in relation to the periphery requires architectural choices to be made. If the assets are high in number and widely dispersed to remote areas, a strong peripheral computation capacity is necessary, especially if latency, local control, and resilience are critical. Functions performing higher-level computations that incorporate data from multiple systems, but do not depend on strict latency requirements, can be performed by a concentrated computational capability. Generally, IIoT systems use a combination of peripheral and concentrated computational patterns. Distributed computational concentration is most resilient, while concentrated computation has the potential for more robust analytics.

主站蜘蛛池模板: 广西| 辽宁省| 昌邑市| 康马县| 洪江市| 苍南县| 贵德县| 文登市| 和硕县| 泰顺县| 沾益县| 东辽县| 靖安县| 西贡区| 安福县| 通江县| 阿拉善右旗| 丁青县| 辽源市| 平度市| 中方县| 镇雄县| 财经| 嘉荫县| 巨野县| 宁明县| 沙雅县| 长寿区| 宕昌县| 仙游县| 商水县| 嘉荫县| 额济纳旗| 辽源市| 唐海县| 凤翔县| 清镇市| 景德镇市| 乌兰察布市| 大埔区| 齐河县|