Co-Authored By:
Sachin Ganu, Distinguished Engineer
Dongwoon Hahn, Lead AirMatch Developer
Jose Tellado, Aruba Fellow and Wi-Fi Analytics Chief Technologist
Artificial Intelligence (AI) and Machine Learning (ML) – a subset of AI – has become the rage in the tech sector. Analysts estimate that there are over 1,000 IT products that claim some AI or ML capability.
In reality, ML comprises a relatively small set of algorithms or models, that learn and make judgments without being explicitly programmed for every situation. When presented with new inputs, the goal of ML is to provide new, intelligent insights such as identifying clusters with common characteristics or predicting a specific outcome.
While using ML can be challenging technically, it is possible that some clever programmer can find a model and send it some data, then call it “Machine Learning.” Industry observers call this “AI washing,” and the result is an overuse of the term “Machine Learning.” ML effectiveness lies not in the algorithm but its application. The difference between marketing hype and successful, AI-based solutions lies in a combination of domain expertise, access to rich training data, scale and real-world exposure.
Training on relevant data is crucial for using ML effectively. 100,000 data points sound impressive, but successful training starts with millions of data elements. Model maturity and effectiveness develop over the years, not months.
Let’s take the example of autonomous networking. Self-learning, self-adjusting Wi-Fi networks rely on ML to diagnose the current environment and make the right changes to optimize the user experience. Relying on newly-minted ML that hasn’t been properly trained and exposed to a wide variety of real-world situations is like asking a teenager who has only navigated the family driveway to face rush hour traffic.
The Aruba AI advantage starts with a deep understanding of Wi-Fi and security technology and a history of data science success commanded by a strong team of data scientists. This is supported by insight and data from thousands of installations ranging from small buildings to campuses to stadiums, across all verticals and geographies, and from small, emerging businesses to large, global deployments.
At the Aruba Atmosphere Conference this week in Bangkok, we’ll showcase the value of Aruba’s experience across a broad range of mature, ML-based solutions that are the foundation of our autonomous networking solutions and advanced cyber-attack detection.
NetInsight collects network operations data to recognize changing usage patterns, predict issues and recommend network changes to improve Wi-Fi network performance for users and IoT devices.
IntroSpect, Aruba’s advanced attack detection solution, spots attacks that have evaded traditional cyber defenses that can only be found by looking for small changes in behavior. IntroSpect user, and network behavior analytics uses both supervised and unsupervised ML for not only advanced attack detection but also to accelerate incident investigations and response.
AirMatch is a capabilit y built into ArubaOS8 that operates at the edge of the network to automatically manage the RF spectrum for the entire network and dynamically adjust RF settings to maximize Wi-Fi stability, performance and predictability without IT intervention.
ClientMatch has been supported since ArubaOS6.x and focuses on the problem of monitoring and proactively maintaining Wi-Fi device connectivity to the radio that will service it best. It addresses challenges with sticky clients, band selection and spectrum load balancing by intelligently updating connections based on device types and how they inter-operate with the network.
Cape Networks monitors network performance from the perspective of a user or device, including application interactions. The user insights from Cape provide effective input to the ML models as well as a source of validation when network changes are made.
The Aruba AI advantage is built upon deep domain expertise that comes from 16 years of wireless and wired networking leadership. By selecting the right problems to address, knowing what data is most relevant, and having access to it in sufficient quantity to train the ML models and then curate the result in the most demanding of real-world environments, Aruba is setting the standard in delivering AI-powered, secure and autonomous networks.
Learn more about Aruba's AI-powered solutions.
