In the past few years, information security has always been based on a combination of anti-virus, isolation and encryption technologies. Government agencies and information security companies are willing to use methods to track Internet traffic and find suspicious materials based on their signatures. The focus of these technologies is to detect malware after a problem has occurred and to isolate the good data from the malware. However, if malware is not detected, it may be lurking for months or even years in the background of the system and become active later.

The consumer sector is changing rapidly. It is migrating from an environment where only computers, game consoles and smartphones are connected to the Internet. This environment integrates new devices such as sensors, cameras and smart appliances, with the goal of giving their owners and users a real-time view of many things in their lives: housing conditions, family affairs, personal safety, weather and more.

Now, we have a more complex environment with more and more devices, each of which can be the target of the attack, and there are privacy and security vulnerabilities. However, in addition to laptops and smartphones, these connected devices can usually perform at most one or two functions. If they originate from a design purpose, the monitoring station can alert the central system and flag the problem. This is the important role that artificial intelligence (AI) and machine learning (ML) can play in protecting the surrounding environment of consumers.

The importance of artificial intelligence and machine learning to protect consumers

Machine learning can be used to determine the behavior patterns of a system, such as traffic on a network, running applications, and communication established between devices. The machine learning system will track patterns in the device, on the local network, or in the cloud.

At the device level, the local machine learning system will determine the normal operating mode of the device by looking at a series of parameters such as memory, tasks, and IP addresses, and determine how it will operate under normal conditions. In smart home appliances with only one or two functions, good modeling of behavioral patterns can be achieved by embedding a neural network accelerator (NNA) that enhances the machine learning engine. Devices can report their metadata to network-level or cloud-level systems, which will receive all of this information and analyze it across a wide range of device groups.

At the network level, routers can view all traffic and can use their intelligence to determine when devices in the network communicate with the outside world. By using a machine learning engine, they can assess when anomalous communication occurs, and can detect anomalous data flow from the network to the outside world, which can be reported as a problem. Vice versa, they can identify anomalous traffic sources for local devices.

In the cloud, application hosts can see a wide range of devices and networks, and with their large computing resources, they can track real-time activity across the entire environment. They apply the same machine learning concepts as the device or network level, but because of their computing power, they can process more data and view more specific information about the vast ecosystem.

Experience from commercial and industrial markets

Machine learning and forensic analysis have become commonplace in industrial and commercial environments. There are successful examples of machine learning-based security technologies in industries such as hospitals, transportation systems, factories, and oil and gas platforms. Machine learning is used in conjunction with traditional techniques for separating sensitive data and tracking known attacks. It provides an additional dimension for early identification of destructive behavior through analysis. Tracking individual devices has become increasingly difficult due to the growing challenges of the connected device ecosystem. The help of an artificial intelligence system is needed to determine when a device is infected with malware.

The machine learning system will be able to detect attacks such as the Mirai botnet caused by malware installed in the webcam. The botnet launched a denial of service (DoS) attack on an Internet directory server on the East Coast of the United States. At the device level or at the network level, attack-related anomalies are detected by using machine learning techniques and the device owner is notified as early as possible.

Artificial Intelligence Information Security in 2020

The application of machine learning in the consumer field is very extensive. From checking whether privacy parameters have been set up and tracked regularly, to observing device operations, protecting consumer data and private information, machine learning systems have become the custodian of the consumer environment. It is placed in the cloud of devices, routers, and hosted applications that work together to provide guidance for setting up devices and protecting consumers.

By transmitting metadata devices and network metadata to cloud-level systems, devices and networks can perform cloud analysis and forensics activities. The Cloud Machine Learning and Analysis System provides a bird’s eye view of a vast ecosystem that connects behavioral patterns across networks. Although these technologies were originally pioneered in the commercial and industrial markets, they are fully applicable to the consumer sector.

All in all, the connection of Internet of Things (IoT) consumer devices increases the attack surface of malware. At the same time, by sharing this metadata with cloud operators, it enables machine learning-based analytics to provide security solutions based on local environmental behavior patterns.