As smart building needs continue to evolve, savvy facility managers are discovering one of the most beneficial automation technologies they can deploy is artificial intelligence (AI). However, compared to familiar IoT technologies, like sensors, IA is more complex, and determining where and how to use it can be confusing. Understanding some fundamental aspects of AI can give facility managers the confidence to know their deployment will return the biggest benefits for their building automation investment.

Moving from IoT to AI

Internet of things (IoT) technology, like sensors and smart devices integrated into commercial buildings can monitor and control an extensive range of parameters, including room temperature, IAQ, lighting levels, and occupancy. Many building owners and facility managers rely on IoT-powered building automation systems to reduce energy use and make spaces healthier and more comfortable for occupants. And while the information IoT-powered BASs provide is valuable, the capabilities of building automation controlled by AI can exponentially increase that value.

Smart buildings can contain hundreds or thousands of IoT devices, and the resulting data streams can be bewildering. It’s too much for even an entire building staff to track and interpret. But AI models can do that and more. AI-controlled BASs can release the full potential of IoT deployment by interpreting a building’s vast amounts of data and providing calculated insights to operators. Over time, AI can learn from that data and, with facility manager oversight and approval, make autonomous decisions and predict the building’s future. This kind of agile, deep-visibility facilities management can help achieve ESG goals and exceed occupant expectations.

Processing and Storing Data

Moving to AI integration includes a range of decisions, but a fundamental one is where data will be processed and stored. Machine-learning algorithms need considerable computing power for delivering insights based on what’s learned. Originally, AI systems were cloud-based, since local infrastructure didn’t have the resources to effectively complete these tasks. However, running AI-powered building automation out of remote data centers has limitations, including connectivity, bandwidth costs, security, and latency. These limitations can result in downtime, high costs, security risks, and data lags, which can influence the efficacy of a system. If there’s an equipment problem or system failure, personnel need to receive automated alarms as quickly as possible so they can address it.

In response to cloud limitations, a new generation of edge computing technology, like that used in the OpenBlue building automation platform from Johnson Controls, has emerged. Through a type of distributed computing and an advanced technology called Edgification, this edge infrastructure has the processing power AI demands and is installed on premises, reducing issues related to reliability, security, and latency as well as costs. For these reasons, edge AI is proving its value and growing in popularity. Gartner estimates that by 2025 75% of all data will be processed at the edge.

Even as many AI-powered BASs move to the edge, the cloud remains valuable. It’s important for facility managers to know when an application calls for edge AI and when it calls for cloud AI.

Choosing the Best Platform

Both cloud and edge AI get their names from where the AI computation occurs. Cloud AI computation happens using the internet (aka cloud) and data center servers. Edge AI happens at the edge of a network, which includes servers that are typically on the premises and close to the smart devices generating data.

Cloud AI and edge AI each have distinct advantages. Here are a few questions to consider when deciding whether cloud, edge AI, or a combination of both would be more advantageous for your smart building automation system applications.

How’s your Internet Connection?

The internet can be a factor outside of your control, especially in the case of an outage. Network disruptions can significantly affect operations. Cloud AI requires an internet connection with high-speed bandwidth because devices need internet access to function and make decisions on large volumes of data. Edge AI, on the other hand, doesn’t need internet access.

How Quickly Do You Need to Access Data?

To operate efficiently and prevent issues, smart automation systems that detect operational problems and automatically respond need to process and access data as quickly as possible. That response time can be affected by both bandwidth and latency.

Measured in bits per second, bandwidth is the frequency of data sent to and received from the cloud. Latency refers to data speed — the time it takes to send information to the cloud and back — and is typically measured in milliseconds. Latency is related to but different from bandwidth: The faster data must travel, the higher bandwidth needs to be. High latency means the lag time is greater, and low latency means there’s little to no lag time.

Generally, cloud AI has lower bandwidth and higher latency than edge AI. Edge AI is favored for its high bandwidth and low latency, especially when actions need to be executed in real-time or close to it.

How Sensitive Is Your Data?

Security is critical when it comes to protecting your data and organization. The local control provided by edge AI can improve privacy and cybersecurity for critical building automation applications. Turning off a machine or adjusting a control system from the cloud can increase security risks and potentially create latency issues.

How Much Information Do You Need to Transfer and Store, and What’s Your Budget?

Data comes in all sizes, and some files are bigger than others. When it’s continuously collected, high-fidelity images, audio, and video — and the costs to transfer and store them — can add up quickly. For example, a commercial building may have a video monitoring system that produces high-fidelity images from multiple cameras that are analyzed by a computer vision AI model. If a system uses cloud AI, costs depend on the amount of data that’s transferred and stored, and that amount of data may quickly become unaffordable. With edge AI, data is typically stored in smart devices or a local server.

Do You Want Deeper Operational Insights?

Rigorous data analysis usually doesn’t need to happen in real-time, but it can require the most powerful hardware and software tools at any scale. Cloud AI may be the best choice if you want to better understand your operations based on AI analytics or run simulation exercises on a digital twin of your facilities.

Do You Manage More than One Property?

If you need to correlate information between properties, cloud AI can allow for a centralized data clearinghouse and command center. However, a hybrid approach is also popular, where initial processing in individual buildings happens through edge AI, then cloud AI is run on the data aggregated from multiple buildings and possibly other data sources.

Deploying for Success

Today’s AI-powered BASs can help make facilities more energy efficient, healthy, autonomous, safe, and responsive to occupant needs — all of which is needed now more than ever. These AI-based systems are a significant step toward a truly smart building that has the intelligence to manage, heal, and secure itself. By deploying AI technology, building managers can lower resource consumption, reduce operating costs, optimize space utilization, increase productivity, improve occupant comfort, gain and retain tenants, and enhance revenue opportunities. Further, AI-powered BASs can help building managers more easily achieve ambitious ESG goals and better accommodate occupants who are back in the office.

While AI started in the cloud, it’s also finding a home at the edge. Edge AI processes and analyzes data where it’s generated, providing near-real-time insights and facilitating localized actions. Meanwhile, cloud AI is still noted for its computation capabilities and vast storage. Each platform can benefit different applications, and a facility manager who knows when to deploy them has a distinct advantage.

To successfully deploy AI, it’s important to partner with a technology vendor who has demonstrated expertise in this field. The ideal partnership will help ensure that when AI is integrated, it will adequately serve the application’s needs and help facility managers better meet the challenges of both today and tomorrow.