The Intelligent Edge
Large scale IoT solutions potentially produce vast amounts of data, from a variety of sources like sensors, equipment and system integrations. One of the key problems is being able to ingest and process the data at the pace it arrives. IoT devices are often constrained by network latency and bandwidth limitations, and further complicating the task, data types are often very diverse.
An approach to optimizing data ingestion is to distribute computation. At the same time edge hardware is becoming increasingly powerful to perform complex processing, including the possibility to aggregate, pre-process, filter or enrich data. Instead of processing all data centrally in the cloud, computation can be distributed such that you bring it closer to the data source – to the edge.
An edge solution has several other advantages over traditional IoT solutions.
Latency and scalability: Processing data at the edge reduces latency since data must travel a shorter distance. Additionally, it provides the opportunity to carefully select what to send back to the cloud. By only sending the data that you really need, you potentially reduce the volume of data drastically, as well as network bandwidth and cloud ingestion resources.
Data captured by edge devices can contain sensitive information, such as video- or microphone streams. By processing data on the edge, it is possible to keep the sensitive data private, by only sending the anonymized result to the cloud. The sensitive data never leaves the device, which means that the risk of security breaches is reduced.
Another feature that edge applications provide is the possibility of autonomous operation. The key to obtain this is to reduce dependencies towards central systems by allowing data processing to run independently from cloud functionality. Machine learning models, tag configurations or other data that is required to perform the processing of data, can be stored directly on the edge device. This ensures continuous operation even during an outage in Internet connectivity. Data integrity is ensured by implementing two-way (up- and downstream) asynchronous communication, where data is buffered in the event of a network outage and forwarded (store-and-forward) as soon as the connection is restored. Autonomous and offline operation results in better uptime, and in the end, improved data quality.
Typical Edge applications could be:
- Data collection / aggregation
- Condition Monitoring (vibration, temperature)
- Decision support systems (AI)
- Video processing / detection
Tricloud provides off-the-shelf frameworks for data collection, time series aggregation and Python script integration, which drastically speeds up development. In addition, Tricloud has expertise in applying machine learning and computer vision technologies.