AIoT for Achieving Sustainable Development Goals

Abstract

Artificial Intelligence of Things (AIoT) is a relatively new concept that involves the merging of Artificial Intelligence (AI) with the Internet of Things (IoT). It has emerged from the realization that Internet of Things networks could be further enhanced if they were also provided with Artificial Intelligence, enhancing the extraction of data and network operation. Prior to AIoT, the Internet of Things would consist of networks of sensors embedded in a physical environment, that collected data and sent them to a remote server. Upon reaching the server, a data analysis would be carried out which normally involved the application of a series of Artificial Intelligence techniques by experts. However, as Internet of Things networks expand in smart cities, this workflow makes optimal operation unfeasible. This is because the data that is captured by IoT is increasing in size continually. Sending such amounts of data to a remote server becomes costly, time-consuming and resource inefficient. Moreover, dependence on a central server means that a server failure, which would be imminent if overloaded with data, would lead to a halt in the operation of the smart service for which the IoT network had been deployed. Thus, decentralizing the operation becomes a crucial element of AIoT. This is done through the Edge Computing paradigm which takes the processing of data to the edge of the network. Artificial Intelligence is found at the edge of the network so that the data may be processed, filtered and analyzed there. It is even possible to equip the edge of the network with the ability to make decisions through the implementation of AI techniques such as Machine Learning. The speed of decision making at the edge of the network means that many social, environmental, industrial and administrative processes may be optimized, as crucial decisions may be taken faster.

Deep Intelligence is a tool that employs disruptive Artificial Intelligence techniques for data analysis i.e., classification, clustering, forecasting, optimization, visualization. Its strength lies in its ability to extract data from virtually any source type. This is a very important feature given the heterogeneity of the data being produced in the world today. Another very important characteristic is its intuitiveness and ability to operate almost autonomously. The user is guided through the process which means that anyone can use it without any knowledge of the technical, technological and mathematical aspects of the processes performed by the platform. This means that the Deepint.net platform integrates functionalities that would normally take years to implement in any sector individually and that would normally require a group of experts in data analysis and related technologies [1-322].

The Deep Intelligence platform can be used to easily operate Edge Computing architectures and IoT networks. The joint characteristics of a well-designed Edge Computing platform (that is, one which brings computing resources to the edge of the network) and of the advanced Deepint.net platform deployed in a cloud environment, mean that high speed, real-time response, effective troubleshooting and management, as well as precise forecasting can be achieved.

Moreover, the low cost of the solution, in combination with the availability of low-cost sensors, devices, Edge Computing hardware, means that deployment becomes a possibility for developing countries, where such solutions are needed most.