AI Gets to Wi-Fi: How Innovations Help Build Wireless Networks in Offices
The logic behind building wireless networks is getting more sophisticated every day. Companies' IT departments must adapt to the changes to keep up with growing demands and expectations of users. Hiring a staff of highly qualified engineers, especially given their scarcity on the market, is not an option for most companies. So what to do? Move to the cloud! Experts believe that, in most cases, the cloud management system will reduce the requirements for the wireless networks experience of the company's staff. Solutions for connecting the latest generations of Wi-Fi in the cloud are based on artificial intelligence and machine learning technologies, which gives companies a completely new way to look at the task of building a corporate wireless network. Gaby Kroet, Mist AI Sales Director—EMEA at Juniper, and Valery Lutoshkin, Head of Network Solutions at STEP LOGIC, spoke about this in detail in an interview with CNews.
CNews: What are the most promising technologies for wireless network infrastructure? How attractive are they for businesses?
Gaby Kroet: The industry is rapidly moving toward cloud-based architectures and microservices for flexibility, resiliency, and elasticity. These scalable architectures also support real-time artificial intelligence (AI) requirements, opening the door to Artificial Intelligence in IT Operations (AIOps) and automation. Now it is possible to configure hundreds of websites in minutes, implement new functionality in weeks (instead of months), troubleshoot problems in advance, and focus on strategic tasks.
Valery Lutoshkin: The construction of corporate networks is in demand, there is a noticeable trend of shifting non-critical services toward wireless networks. For example, laptop manufacturers have long refused to install network ports in their devices. Wi-Fi is often the primary medium for connecting devices to the corporate network. And while servers will most likely never connect to Wi-Fi, workstations, smartphones, and other devices have long been functioning wirelessly.
CNews: What trends related to the mentioned technologies can you identify on the global and Russian markets?
Gaby Kroet: The industry is rapidly transitioning to cloud-based architectures and microservices.
Businesses are all too familiar with the challenges that remote working environment poses to networks and network teams. Over the past year, many companies have had increasing problems with bandwidth and connectivity. Due to the circumstances, they had to fix these problems on their own, although the same process could have been much easier with appropriate AI strategies and automation.
AI can transform the traditional enterprise experience through proactive notifications, automation, smart analytics, and streamlined help desk processes. The goal of the AIOps solution is to provide ease of use through automation and ultimately create a better IT experience for both internal staff and customers.
In addition to investing in AI, it is critical for enterprises and IT executives to develop a robust cloud model for their networks. It must include the capabilities of software-defined wide-area networks.
As a result of the need to support remote operation and provide secure and reliable branch office connectivity, SD-WAN (Software Defined Wide Area Network) technology has become critical to IT teams. SD-WAN with AI is an advanced innovation, a useful IT management tool, because it provides full visibility of what's going on in the network. This solution prevents and automatically fixes problems.
Valery Lutoshkin: During the years of rapid innovation, the quality of wireless networks was lagging far behind the needs of any growing business. The available technology could not provide the level of service that companies needed. A classic problem: data loss and unpredictable delays (so-called "jitter") occur in a newly built wireless network, resulting in poor quality of critical systems, such as IP telephony. That's why, until recently, wireless networks could not be used for services with guaranteed operations where a specific level of data quality was necessary. In recent years, developers have paid special attention to this problem, and the new wireless access technologies Wi-Fi 6 (aka 802.11ax) and 5G include solutions to provide customers with a guaranteed level of service.
Wi-Fi 6, the latest generation Wi-Fi standard, successfully solves the issue of customer density per access point. Since the network is one, and there is no way to build a separate isolated communication channel to each device, the Wi-Fi network works in a shared environment. According to the tests, the equipment of Wi-Fi 5 standard allowed to connect up to 100 users to one access point with the speed up to 2 Mbps each, although in theory, the standard provides much higher speeds (the maximum theoretical speed—about 7 Gbps). Such a serious limitation on the number of simultaneous users has been overcome with the advent of Wi-Fi 6. Under the same conditions, the Wi-Fi 6 access point provides up to 400 clients with the same bandwidth (with a maximum theoretical speed of about 9.6 Gbps). Although, of course, it should be noted that the best performance is achieved when each of the devices used by customers supports Wi-Fi 6.
CNews: What role do network solution providers assign to cloud AI in their developments? What problems can be solved with this technology and how does it relate to the personnel qualifications?
Gaby Kroet: Organizations across all industries are undergoing digital transformation that has only accelerated because of the pandemic. This poses another universal corporate problem for IT teams: as budgets shrink, the need to do more with less grows. The focus is on three cornerstone technologies: AI, machine learning, and automation. Consequently, there is a growing need to move architectures to the cloud.
The cloud is the answer to IT's desire for best-priced, high-quality solutions that deliver secure, consistent, and reliable operations. Using AI helps improve operational efficiency and customer satisfaction. AI also allows businesses to better align with business goals and ensure the outcomes.
Valery Lutoshkin: Any company that has a wireless network in its office faces the task of monitoring events, tracking incidents, and finding ways to eliminate them, preferably with minimal staff involvement. Over the past decades, manufacturers have been approaching such challenges from different angles. Previously, there has been an attempt to collect log files from all devices in the network, that is, to record the events occurring, correlate them with each other, and draw conclusions. Today's solutions use artificial intelligence technology instead of simple correlation, which builds a chain of patterns based not on engineers' predictions but on real events. The cloud-based AI system captures all the events that occur on the devices, analyzes variants of their development and potential consequences. AI has a self-learning mechanism, so surely the neural network gets smarter over time.
When a company decides to maintain a wireless network in-house, there is a need for highly skilled engineers. But such professionals are few and expensive. Companies that don't find specialists face problems in the quality maintenance of corporate systems. That's why vendors are proposing to bring this part of analytics to the cloud, which will cover the needs of the company without the pressure of hiring networking professionals.
CNews: Tell us about Juniper's Mist AI technology, please. What tasks does it help to solve and what are the characteristics?
Gaby Kroet: Mist AI's fundamental mission is to assist companies in building self-managing networks using machine learning technology, advanced big data analytics science, and automation of operations.
Juniper's Mist AI-powered wireless services are designed to make network performance predictable, reliable, and measurable, delivering a transparent user experience. The solution performs AI-powered automation and, through self-recovery capabilities, reduces network operating costs, eliminates routine and the human factor.
Juniper access points collect data and implement policies in tandem with Juniper Mist cloud, it's an essential environment for successful analytics, machine learning, location services, and event correlation functionality. This creates a single corporate-level platform for Wi-Fi, Bluetooth LE, and the Internet of Things.
Artificial intelligence technology makes it possible to create digital assistants, or chatbots, in any industry. Juniper was the first in the industry to create a virtual network assistant (VNA) named Marvis. It acts as a digital expert, supporting network professionals—administrators, reliability engineers. Marvis is continuously enriching its Natural Language Processing (NLP) and Language Generation (NLG) capabilities to offer a conversational interface that understands user intent and provides increased value and quality of the results returned. The convenience of the Marvis dialog interface is that it can contextualize queries for different purposes: to speed up workflows, troubleshoot problems, make intelligent decisions or recommendations to get real-time answers about the network, and to improve a specific user experience by learning from user feedback.
CNews: What are the technological features of AP43 and what are its advantages over other access points?
Valery Lutoshkin: The AP43 solution makes it possible to build a completely wireless network in the office. First, it increases the density of coverage per point, so you can get by with installing fewer points in a room without losing data transmission quality. The second advantage is reduced lagging, so interactive applications will become more stable and responsive from the user's point of view. The third advantage is optimization of power consumption for mobile devices. The batteries of many laptops and smartphones can be completely drained within a working day. The special TWT mode in Wi-Fi 6 allows you to reduce power consumption by about a third. Consequently, a device connected to the Wi-Fi 6 network will last longer.
CNews: How is AP43 used in the existing Wi-Fi networks? What serious measurable business benefits does this solution bring?
Valery Lutoshkin: Juniper Mist Wi-Fi system can be used in various industries. For example, to run targeted ads based on the customer location in retail, to analyze data on human contact in certain spaces during the COVID-19 pandemic or the location of staff near patients together with temperature and humidity monitoring in medical institutions, to solve the navigation problems in airports, university campuses, entertainment centers, and to inform users about the occupancy of restaurants.
The main advantage for business is the reduction of operating costs. Customers with large installations (more than 10,000 access points) who have replaced their infrastructure with Juniper Wi-Fi Mist estimate a cost reduction of more than 40%.
This is due to the lack of need to operate a complex infrastructure of Wi-Fi controllers in their DPCs, and the simplification of network deployment and radio planning has resulted in significant time savings in troubleshooting, full automation of software updates, and maintenance of access points—up to the automatic creation of return merchandise authorizations (RMA) in case of equipment failure.
One of the great advantages of the Mist solution is that it is cloud-based, which means that it is very easy to install. Once the point has been acquired, the owner registers a service contract on the Juniper website and receives a login and password to the cloud. The equipment connects to the cloud by itself and links to the owner's account using the serial number. The interface displays an access point on which pre-installed in the cloud services can be launched as needed. At the same time, most of the services, such as the ability to track customer locations, are available by default and ready to go. Thought-out, ready-to-launch functionality greatly facilitates the work of companies that do not have highly skilled personnel to configure and maintain Wi-Fi network. Anyone with a basic understanding of wireless networks can set up such a system.
CNews: How is the system maintained and monitored? Is the automatic control mode available?
Valery Lutoshkin: The system applies integral service quality assessment based on SLE (Service-Level Expectations) metrics. This approach allows real-time monitoring of service quality by such criteria as connection time, connection success, network coverage, roaming efficiency, bandwidth, radiofrequency capacity, and access point status. Each metric consists of many parameters and has its own individual thresholds for triggering—they allow you to inform the owner of Wi-Fi infrastructure of potential or current problems in advance.
The entire monitoring process is automatic, but also the administrator can get information about the events for each user, including historical data.
In addition to wireless SLE metrics, Mist AI includes SLE metrics to monitor the campus wired network as well as SD-WAN metrics.
CNews: Is there a mechanism for identifying network issues or overloads? How does it work?
Valery Lutoshkin: Malfunction detection is also based on SLE metrics. They are generated from telemetry data coming from access points.
There can be many reasons for wireless network overload, and each case has its own metric: high bandwidth utilization, presence of Wi-Fi interference, presence of non-Wi-Fi interference, clients that create high load, problems with network equipment, limited coverage, incompatibility of user devices.
Any difficulties that arise, such as the emergence of illegitimate access points, targeted DoS attacks, or problems in the wired network infrastructure and their services, are covered by their own SLE metrics and displayed in the Mist AI system.
The access point has a mechanism for dynamically capturing Wi-Fi packets in case of user anomalies, which significantly reduces the time spent on troubleshooting. For example, you can get traffic analysis of the user authorization process in less than one minute and understand the root cause of the failure. The troubleshooting procedure is greatly simplified by the previously mentioned Marvis, an intelligent assistant that works off human language queries or its own query language.
Another sensitive aspect for companies is the ability to track illegal devices within the wireless network's range. AP43 performs continuous monitoring of the radio frequency spectrum while providing users with a basic service. No matter what band the access point is configured in, the security sensor detects rogue access points (Rogue, Neighbor, Honeypot) in the 2.4 and 5 GHz bands, localizes them, and reports the incident to the administrator.
CNews: How do you see the future of this solution in Russia?
Valery Lutoshkin: A generation change of IT specialists is now underway in Russia. Professionals with years of experience are often wary of cloud solutions because of fears that the cloud will cease to be available and the service will suddenly disappear. The new generation, on the other hand, lives in the era of cloud solutions. Young professionals appreciate all the benefits of the cloud and are embracing cloud-enabled wireless solutions. An important fact about the cloud that everyone recognizes: in terms of quality to price ratio, cloud solutions are unrivaled. They allow you to invest less and get quality service for a subscription fee. There is a tendency for such solutions to appear on the market, and it is progressing. According to our estimates, in five years, the majority of wireless solutions in the SMB segment will be built using cloud technology.
Author: Anna Malygina