Mist automates WLAN monitoring with new AI features

Mist says its AI platform for WLAN monitoring can now handle hundreds of queries and will use deep learning to flag anomalies before they cause problems for users.

Mist Systems announced this week that its Marvis virtual network assistant now understands how to respond to hundreds of inquiries related to wireless LAN performance. And, in some cases, it can detect anomalies in those networks before they cause problems for end users.

IT administrators can ask Marvis questions about the performance of wireless networks -- and the devices connected to it -- using natural language commands, such as, "What's wrong with John's laptop?" The vendor said the technology helps customers identify client-level problems, rather than just network-wide trends.

Marvis could only handle roughly a dozen basic questions at launch in February. But Mist's machine learning platform has used data from customers that have started using the product to improve Marvis' natural language processing (NLP) skills for WLAN monitoring. Marvis can now field hundreds of queries, with less specificity required in asking each question.

Mist also announced an anomaly detection feature for Marvis that uses deep learning to determine when a wireless network is starting to behave abnormally, potentially flagging issues before they happen. Using the product's APIs, IT departments can integrate Marvis with their help desk software to set up automatic alerts.

Mist has a robust platform for network management, and the advancements announced this week represent "solid steps forward for the company and the industry," said Brandon Butler, analyst at IDC.

Cisco and Aruba Networks, a subsidiary of Hewlett Packard Enterprise, have also been investing in new technologies for automated WLAN monitoring and management, Butler said.

"Mist has taken a unique approach in the market with its focusing on NLP capabilities to provide users an intuitive way of interfacing with the management platform," Butler said. "It is one of many companies ... that are building up their anomaly detection and auto-remediation capabilities using machine learning capabilities."

Applying AI to radio resource management

The original promise of radio resource management (RRM), which has been around for 15 years, was the service would detect noise and interference in wireless networks and adjust access points and channels accordingly, said Jeff Aaron, vice president of marketing at Mist, based in Cupertino, Calif.

"The problem is it's never really worked that way," Aaron said. "RRM has never been real-time; it's usually done at night, because it doesn't really have the level of data you need to make the decision."

Now, Mist has revamped its RRM service using AI, so it can monitor the coverage, capacity, throughput and performance of Wi-Fi networks on a per-user basis. The service makes automatic changes and quantifies what impact -- positive or negative -- those changes have on end users.

Mist has RRM in its flagship product for WLAN monitoring and management, Wi-Fi Assurance.

Service-level expectations for WAN performance

Mist will now let customers establish and enforce service-level expectations (SLEs) for WAN performance. The agreements will help Mist customers track the impact of latency, jitter and packet loss on end users.

The release of SLEs for the WAN comes as Mist pursues partnerships with Juniper and VMware to reduce friction between the performance and user experience of the WLAN and the WAN.

Mist also lets customers set service levels for Wi-Fi performance based on metrics that include capacity, coverage, throughput, latency, access point uptime and roaming.

Dig Deeper on Network management and monitoring

Unified Communications
Mobile Computing
Data Center
ITChannel
Close