Whitepapers

Edge computing and AI, a winning pair: the pluses for data analysis

Cloud computing allows for significant cost cutting in terms of data centers, providing access to resources for storage and calculating capacity that are virtually unlimited and available on-demand. But not all applications are suitable to cloud solutions, which have various potential drawbacks connected to latency and connection reliability. In these cases, a far more preferable solution is edge computing, the system capable of processing information on the device where it’s generated, eliminating the problem of both data transfers and latency. This is even more true when edge computing is combined with AI.

 

What is Edge AI, the combination edge computing and AI

When edge computing embraces artificial intelligence (AI), the result is Edge AI, permitting real-time data processing without the need to connect to a cloud.

 

Connecting to a cloud, in fact, is not always possible, but, in the case of self-driving (or driverless) cars and other AI applications that require immediate reactions regardless of connection status, Edge AI represents a viable option. The benefits of edge computing are visible in the health care field, for example, where telemedical services are steadily gaining ground and where the need to access data as fast as possible can truly make the difference.

 

Edge computing and AI: the advantages of a winning pair

Edge computing with AI becomes the best approach in the following cases:

 

·         when internet bandwidth is insufficient or insufficiently reliable;

·         if the connection to send data to the cloud isn’t robust;

·         when extremely rapid response times or specific security and privacy needs regarding data transmission are required.

 

Edge AI has the same advantages as edge computing: speed, less network consumption, security, privacy

 

The greatest benefit of the combination of edge computing and AI lies in the ability to drastically reduce the periods of inactivity in cloud-based AI processes, offering users the opportunity to manage them from smart devices capable of reacting quickly to the input without transmitting data elsewhere.

 

The reduction of latency, so crucial in time-sensitive applications, is such that when AI calculations are performed at a remote data center, the best-case scenario envisages a latency reduction of at least 1-2 milliseconds (if not hundreds).

 

Edge AI will also be strengthened significantly by the growing spread of 5G connectivity. Compared to LTE, 5G will ensure higher speeds (20 Gbps versus 1 Gbps), simultaneity of multiple connections (1,000,000 per km2 versus 100,000) and a higher speed of latency (1 ms versus 10 ms). With the gradual adoption of 5G, new-generation connectivity will drive the development of the IoT market and of Edge AI because it facilitates interactions between a wider range of devices, increasingly equipped with smart processors.

 

Edge Analytics: how it changes data analysis for business

By 2022, the Observers of the Polytechnic University of Milan estimate that every home will have an average of 50 devices connected to the internet: the amount of Big Data available will explode. Data from smart sensors and machines, from social media, data generated by IoT devices: Edge Analytics helps alleviate traffic on the geographical networks that connect the data center to the “edge” of the company network, where data are being generated more and more frequently.

 

Edge Analytics allows a business to analyze an enormous amount of data in real time to improve decision-making processes, aggregating the calculating capacity in network devices (such as switches), peripherals, devices and sensors or ultra-compact micro-data centers, thus redefining the approach to IoT. This new approach is called IoT Edge, boosting the intelligence and analytical capabilities of peripheral devices to improve the performance and security of IOT architecture. In IOT devices the cloud is also streamlined, saving network use for storage and data analysis, and drastically reducing the use of CPUs, GPUs and memory.

 

The market recognizes this potential. According to Statista’s data, in 2019, the industrial edge computing market was worth $11.56 billion. It will continue to grow at an average rate of 18% (CAGR) until 2025, when it’s estimated to reach $30.75 billion. The global market for Edge AI software alone will reach $1.835 billion by 2026, registering a compound annual growth rate (CAGR) of 20.8%. Driving the market will be the increase in businesses’ cloud-based workloads and the rapid growth in the number of connected devices (including wearable tech, growing in the post-pandemic period) and smart applications.

 

Greater security and privacy protection with edge computing and AI

When we talk about business data, another unquestionable advantage of Edge Ai is privacy. From both a legal and technical opportunity standpoint, particularly linked to privacy and to information transmission speed, edge computing will assume a key role in the technologies of the digital European Union. It respects, in fact, the principles of Privacy by Design and Default of the European Data Protection Board (EDPB), especially with regard to smart cars. From the standpoint of cybersecurity, AI processing on a local device prevents you from having to transfer information to another device for data analysis. This significantly reduces the risk of interception.