Mist computing

Mist computing takes Fog Computing concepts further by pushing some of the computation to the very edge of the network, to the sensor and actuator devices that make up the network. With Mist computing the computation is performed at the edge of the network in the microcontrollers of the embedded nodes. The Mist computing paradigm decreases latency and increases the autonomy of a solution.

Fog and Mist (and Cloud) computing are complementary to each other – the application tasks, which are more computationally intensive can be executed in the gateway (the Fog) while the less computationally intensive tasks can be executed in the end devices (the Mist). The collected and processed data is still delivered to the Cloud, where it is made available to the user. An application in Mist computing is a collection of services, which can be distributed among the embedded computing nodes.

Mist computing decreases latency and increases the autonomy of a solution.

Fog Computing was introduced by Cisco because it became clear that the billions of IoT devices that will be deployed over time cannot operate by simply having connectivity to servers, instead the computation is pushed closer to the edge of the network – to the gateways. However, there is an opportunity to further distribute the computation tasks among the network nodes, based on their capabilities and roles any given application.

In Fog computing the gateway bears the responsibility for IoT application execution, regardless whether the application is just simple data collection or building automation with many actuation tasks. Moving the application logic to a gateway has many advantages, such as simple management of application logic (the application logic is all concentrated to a single device) and having access to macro-level application information as data from all nodes is collected to the gateway.

Cloud computing has an important role in IoT as the application is typically made available to the users via the Cloud and the data collected by the IoT devices resides in the Cloud.

However, while it is inevitable that part of the application logic also resides in the Cloud, it is not optimal to move the entire IoT application to the Cloud. Maintaining some of the logic at the edge off the network reduces substantially bandwidth requirements and reduces application latency.