Edge Intelligence: The Central Cloud is Dead – Long Live the Edge Cloud!
alarm should that behavior change due to accident, attack etc. Since decision-making is local, the transmission bandwidth would be very low and latency could be in seconds. The intelligent gateway in this case would function only as a data consolidator and could also have UI, etc., installed.
Physical Access Control – Tailgating Detection at Doors
Tailgating at security doors can be achieved by combining MI module and an inexpensive stereo camera, to ensure that only the right person got through secured doors. Badge readers( BLE / RFID for example) can add additional security,: i. e., the door opens as the person approaches, but only if a valid signal is read from each person( if there is more than one) approaching. If more than one person approaches, and fewer badge signals are read, then an alarm will be registered if they all enter the doorway. By using an inexpensive camera with a processing module, latency and bandwidth are reduced over sending to the cloud.
Fire Detection via Surveillance Cameras
An MI module is used with access to video streams incoming from non-specialized video cameras and scans those streams for fire traces using a trained model. In the case of a fire, latency is an issue, and the local processing can greatly increase response to minimize loss. When there are no incidents, bandwidth is minimized and only summary data is sent.
TECHNOLOGY GAPS intelligence that need to be addressed to support requirements for the use described in the previous section.
Factory Productivity Improvement
( 1) Credibility of information generated by edge computing
In order to realize edge intelligence, it is necessary to establish credible data, as the data are the raw materials for future analysis and decision-making. Now, there are technologies which can be used to ensure the credibility of data exist, such as technologies to secure the identity and the integrity of device-generating data and technologies to protect information being communicated over a network. In addition to enhancing technologies, systematic ways to guarantee credibility of the data by making use of such technologies and operational measures including physical security.
( 2) Assisted / automatic |
optimization |
of |
system operation |
|
|
In order to reduce the operational cost, it is necessary to optimize system operation and to adjust operational parameters more efficiently. At present, many productivity decisions in a factory are done manually by experienced engineers, and even the data for such decision-making can be collected. It would be desirable to make optimization easier by automating data collection and assisting engineers in the optimization analysis, ultimately automating optimization processes.
This section describes different technological gaps related to edge
IIC Journal of Innovation- 25-