Integrating AI and the Internet of Things

The IoT is a network of interconnected devices which communicate with each other and other systems via the Internet, enabling process automation and remote management. Devices like sensors, thermostats and household appliances send data to central systems. There, the data are analysed and then used to improve communication between people and the things. The IoT is used in numerous fields, including smart homes, industry, transport, health care and agriculture, contributing to greater efficacy and user comfort.

Integrating the IoT with AI increases its capabilities. Unlike a traditional IoT, which is primarily focused on connectivity and basic data collection, an integrated IoT uses AI for real-time data processing, predictive analysis and faster decision making.

The IoT uses federated learning, an approach where machine-learning models are trained on data distributed among multiple devices or locations without the need to send the data to a central server. Instead, they remain on the users’ devices and only updates such as neural network weights are modelled and sent to a central server, where they are collected to create a global model. The most problematic issues with the IoT stem from personal data protection. However, federated learning increases data security thanks to the fact that nothing but data updates are shared.

As I noted at the beginning of this article, integrating the IoT and AI is changing how vital sectors of the economy operate. Once integrated, the power they forge together can be leveraged in numerous areas, as I shall show in the following sections of the article. The statistics provided are all drawn from “Exploring the Role of Artificial Intelligence in Internet of Things Systems: A Systematic Mapping Study” by Umair Khadam, Paul Davidsson and Romina Spalazzese.

IoT, AI and big data technologies are paving the way for remote diagnosis and treatment. This is triggering a major transformation of the health care system, heightening both the quality and the efficacy of medical services. AI-equipped IoT devices like wearable sensors, for example, support diagnosis, the automation of routine tasks and data-driven decision-making. 

One of the key aspects for AI in health care is pattern recognition. This includes event recognition (50%), authentication (32%) and diagnosis (14%). AI also plays a vital role in data management, encompassing the reduction of data noise (42%), removal of sensitive information (33%) and filling in missing data (25%). Integrating the IoT and AI not only facilitates more effective monitoring of patients’ health, but also introduces more advanced diagnostic support systems. This, in turn, translates into better treatment outcomes and medical facility management.

Agriculture is another field where IoT and AI integration is contributing to significant improvements in efficiency. Here, AI supports decisions about the allocation of resources (64%), oversight (27%) and planning (9%). AI systems integrated with the IoT facilitate soil management, including protecting the soil and appropriate methods for preparing it. They also make it possible to monitor crop condition, detect diseases and combat weeds.

Operational decisions are the primary aspect of AI support in agriculture, accounting for as much as 91%, with strategic decisions playing a lesser role (9%). Data management in agriculture covers the removal of sensitive data (38%), filling in missing data (37%) and reducing data noise (25%). Pattern recognition such as event detection (50%), authentication (33%) and object recognition (17%) help with monitoring weather conditions and counter threats to crops. Integrating AI and the IoT facilitates smart resource management and minimises losses, making agriculture more sustainable and profitable.

Energy is the third field where AI and the IoT play a major role in managing and improving operational effectiveness. The main task for AI is to support decision-making and this is focused on every decision pertaining to operational matters (100%). Pattern recognition consists of event detection (71%) and authentication (29%), providing assistance in monitoring grids and identifying anomalies. In this sector, resource allocation is also vital, with the use of integrated AI and IoT at more than 80%.

The fourth major field is industry, where the IoT and AI are changing the way production and logistics processes are managed. In this sector, AI is tasked primarily with pattern recognition, which consists of event recognition (27%), diagnosis (27%), authentication (27%) and object recognition (13%). 

Decision support is the second most frequent use for AI in industry, with operational decisions predominating over strategic, at 86% and 14%, respectively. Data management in the form of filling in missing data (43%), data noise reduction (29%) and removing sensitive data (28%), is pivotal when it comes to better information management and process optimisation.

Transport is the fifth primary field for the use of the IoT and AI. Here, the top task for AI is pattern recognition, composed of event recognition (40%), authentication (40%) and diagnosis (20%). Decisions relating to resource allocation (75%) and control (25%) are vital to areas like traffic management, route optimisation and, probably most crucial of all, improving safety.

Integrating AI and the IoT in transport makes it possible to respond dynamically to changing conditions en route and it also facilitates fleet management. Data management in this field encompasses data noise reduction (33%), removal of sensitive information (33%) and filling in missing data (33%). 

The integration of the IoT and AI in transport is facilitating the development of smart traffic management systems and autonomous vehicles, contributing to increased efficiency and safety in the sector.

The development of cloud and edge computing technology has enabled the processing and analysis of the large quantities of data generated by the IoT, solving numerous problems connected with the massive volume of information. Integrating these technologies not only means that data can be processed in real time and accurately analysed, but also responds to the challenges associated with vast data transfers, latency and computing power requirements.

Cloud computing provides a centralised resource for storing, processing and analysing the enormous quantities of data generated by IoT devices. It is particularly useful in instances when large data sets are created, such as with smart cities, the industrial Internet of things (IIoT) and health care systems. Integrating AI and the cloud environment boosts the potential of the IoT in a number of ways, as follows:

• scaling models; cloud platforms facilitate the deployment of large and complex AI models with the capability of processing data from multiple IoT devices simultaneously;
• data aggregation; central cloud systems collect distributed data from IoT devices and then apply advanced analytics and machine-learning algorithms to draw conclusions;
• remote access; cloud solutions provide global access to IoT systems;
• cost savings; cloud computing leverages shared computing resources, reducing the need for costly on-premise hardware.

Edge computing complements cloud computing, bringing date processing closer to IoT devices, which reduces the necessity of transferring large quantities of data to the cloud. This approach is suitable in situations where a company is interested in low latency, increased data security in terms of privacy and limited network bandwidth. The advantages of edge computing include:

• real- time processing; edge devices can process data locally, enabling immediate response any time rapid action is required. This is why edge technology is used in autonomous vehicles and industrial automation;
• privacy and security; sensitive data can be analysed locally, minimising the risk of breaches during transfer;
• reduction in latency; by eliminating the need to send data to remote cloud servers, edge computing provides faster response and data transfer;
• energy savings; edge computing reduces energy consumption by cutting down on data transmission. 

Thanks to advanced data analytics, artificial intelligence enables the IoT to learn and adapt to detected patterns, making it more effective and functional. To give one example, in an article entitled “Navigating the nexus of AI and IoT”, Agostino Marengo cites A. Kannammal and S. Chandia, noting that “in smart cities, AI-driven IoT systems allow the analysis of data streams from various sensors to improve urban planning, traffic management and resource allocation”. In the healthcare sector, integrating AI and the IoT facilitates the detection of patterns in patient data, leading to the creation of more tailored treatment plans and the prediction of potential health problems.

The integration of AI and the IoT translates into better performance and wiser decision-making. It is therefore hardly surprising that interest in this solution is surging. The global IoT-AI market, which was worth USD 10.3 million in 2022, is projected to reach USD 91.7 billion by 2032, with an average annual growth rate of 24.8%. 

State-of-the-art data analytics, particularly AI-based, is completely changing the way that data in IoT systems is collected and interpreted. These improvements have made it possible to introduce real-time analyses and predict future events, which has significantly extended the potential uses of large data sets. 

Morengo cites a study conducted by Salah Uddin et al., which provided a practical illustration of IoT-AI capabilities in terms of predicting events. The researchers created a smart indoor agricultural system and their research proved that predictive analysis in IOT can effectively support resource management.

IoT devices are interconnected, meaning that any problem with privacy protection could bring consequences in its wake. This makes the right security measures a critical aspect of an AI-IoT environment. Privacy protection involves securing data against unauthorised access and it encompasses ethical and legal questions concerning the ownership and use of data. Finding the balance between the use of data by AI applications and the protection of users’ data is a challenge because building users’ trust is crucial.

Integrating the Internet of things (IoT) and artificial intelligence (AI) is transforming the way that various sectors of the economy function. The use of advanced analytics and real-time processing means that integrating AI and IoT facilitates more effective resource management, better diagnostics and improved process automation. Cloud and edge computing technologies can support the integration, allowing large quantities of data to be processed with minimal latency. Sectors like healthcare, agriculture, energy and transport are already leveraging the potential of this integration, which is translating into more sustainable development and boosting performance. However, the greatest challenge is privacy protection and data security, which should be remembered while designing and deploying systems grounded in the integration of AI and the IoT.

1. Davidsson P., Khadam U., Spalazzese R.. Exploring the Role of Artificial Intelligence in Internet of Things Systems: A Systematic Mapping Study, “Sensors” Volume 24, Issue 20; 
2. Chandia S., Kannammal A.. Applications of AI and IoT for smart cities, cited in Agostino Marengo, Navigating the nexus of AI and IoT: A comprehensive review of data analytics and privacy paradigms, “Internet of Things” Volume 27, October 2024; 
3. Allayear S.M., Asaduzzaman M., Farzana R., Samaun Hasan M., Salah Uddin M., Rahman M., , Implementation of smart indoor agriculture system and predictive analysis, cited in Morengo, ibid.
4. Oracle, Internet of Things.

Translated from the Polish by Caryl Swift