Intelligence and Cognitive Computing at the Edge for IoT: Architecture, Challenges, and Applications

Abstract

Cognitive computing is the development of computerized models to mimic human behavior. The best examples are virtual assistants such as Siri, Alexa, Cortana, etc. Cognitive computing and AI play a big role in solving problems and building applications to support several domains in Internet of Things (IoT). The downside to AI and cognitive computing is the complexity of the architecture involved in building models that supports IoT. Toward the late 1960s, researchers came up with a technology called cloud computing that allowed individuals and organizations to make use of services provided by the cloud to solve problems. This meant that one no longer needed to worry about the software and hardware components required to develop intelligent models. Although cloud computing has been able to solve most complexity issues, it has several drawbacks of its own – latency, privacy, security, reliability, etc. Domains like healthcare rely on intelligent models to make important time-sensitive decisions and cannot wait for data to be shipped all the way to the cloud to get results. The need for quick and real-time results is what probed researchers to develop Edge Intelligence or Cognitive Computing at the Edge with IoT application services. Edge Intelligence is the deployment of intelligent computerized models close to the data source – on the edge node of an IoT network or very near to the network edge. This chapter highlights and examines the principles, architecture, challenges, applications, and existing industrial implementations of Edge Intelligence.