Containers

Cognitive Systems Corp.’s Amazon EKS journey: Wi-Fi Sensing technology

This post was coauthored by Mike Ravkine, Sr. Director at Cognitive Systems

Introduction

Cognitive Systems Corp., which provides pioneering Wi-Fi sensing technology, WiFi Motion, has redefined how people interact with wireless networks by using existing Wi-Fi signals to transform connected devices into motion sensors.

Cognitive System’s Wi-Fi Sensing technology constitutes a comprehensive software stack that can be effortlessly deployed onto an access point or client device, which enables motion sensing for a wide array of applications across any Wi-Fi-enabled environment. This technology operates seamlessly without the need for additional hardware, utilizing sophisticated algorithms for channel state information analysis, motion pattern detection, and real-time signal interpretation. Potential applications for this motion sensing technology include home monitoring, eldercare, and smart home automation.

With Cognitive Systems’ widespread deployment and operational reach, particularly in densely interconnected environments where high reliability and low latency are imperative for a multitude of applications, the integration of WiFi Motion with an efficient and scalable container orchestration service emerged as an absolute imperative.

The evolution from a single server

In the footsteps of many budding start-ups, Cognitive started their journey on a smaller scale using a single server. However, as their Wi-Fi Sensing solution gained traction, the surge in deployments gradually escalated demand on this solitary server, necessitating a scalable solution. This led Cognitive to adopt Kubernetes—a robust system capable of efficiently managing multiple servers.

However, as deployments scaled, Cognitive encountered challenges that proved far from simple, going beyond basic computer troubleshooting. This led to countless hours spent on the phone with human customer support agents, and Cognitive knew there must be a better option.

The quest for a more responsive partner

As the complexities of managing their growing Wi-Fi sensing technology grew, so did Cognitive’s demand for swift and effective customer support. Recognizing the need for a more robust solution, Cognitive explored cloud providers. In this pursuit, Amazon Web Services (AWS) emerged as a standout candidate. What sets AWS apart is not only its cutting-edge infrastructure but also its human-centric approach to customer support—a pivotal shift from their previous solution as Cognitive now had knowledgeable professionals at their disposal capable of promptly addressing issues and aiding in troubleshooting.

In 2019, Cognitive started their journey on AWS’s newly launched Amazon Elastic Kubernetes Service (Amazon EKS). The frequency of updates, global reach over multiple AWS Regions, and resource efficiency underscored the potential for ongoing growth and innovation within the platform. Transitioning to Amazon EKS therefore seemed natural to secure a reliable, scalable, and well-managed future for WiFi Motion.

Global expansion and latency woes

When Cognitive’s sales team expanded their focus to Europe and Asia, a fresh set of challenges emerged. To illustrate the potential applications of WiFi Motion to their service provider clients, Cognitive created two sample mobile phone application offerings—Home Aware and Caregiver Aware—to showcase how a service provider could seamlessly integrate Wi-Fi sensing into their offerings. Both applications centered around a key feature named Live Motion, which enables users to visualize ongoing motion within their homes through the application in real time.

The effectiveness of this feature relied heavily on low latency, specifically under 250 ms, to guarantee smooth functionality. Moreover, seamless navigation through the prototype app was paramount, particularly since these test apps were frequently employed in service providers’ Wi-Fi sensing trials.

While the latency between North America and Europe maintained a tolerable margin at around 200 ms, Asia posed a far more substantial obstacle. Given the vast geographical expanse, latency consistently surpassed acceptable thresholds, resulting in delays of over a second. Regrettably, this disparity led to adverse user experiences. This led Cognitive to explore AWS’ vast regional presence to provide low latency and improve the user experience.

Using Amazon EKS: A game-changing AWS Region in Tokyo

Amazon EKS’s global reach across multiple AWS Regions played a pivotal role in addressing these latency challenges. With the availability of an AWS Region in Tokyo, Cognitive effectively extended its presence into Asia. The results were remarkable: sample application evaluators, who had previously encountered performance issues, were now enjoying seamless interactions. Amazon EKS was instrumental in turning Cognitive’s vision of a worldwide revolution in Wi-Fi innovation into reality, which made motion sensing accessible to any Wi-Fi connected space. This global accessibility is crucial for Wi-Fi sensing to become the next transformative advancement for the ubiquitous technology of Wi-Fi, which has long been synonymous with communication.

Choosing to standardize on AWS and Amazon EKS yielded benefits beyond addressing latency concerns. The abundant talent pool familiar with AWS facilitated easier hiring while relying on off-the-shelf components of Kubernetes lightened our workload. With a small team, Cognitive was able to efficiently manage 10 Amazon EKS clusters (and counting) today. This efficiency was further highlighted by the absence of live pager duty; the cloud’s inherent self-healing capabilities ensured continuous system operation.

Amazon EKS empowered Cognitive by providing the foundational elements to develop self-monitoring and self-recovery functions, minimizing the need for extensive engineering resources. This approach, combined with AWS’s built-in healing mechanisms like Managed Node groups, mitigated the impact of sporadic server and network failures inevitable at our global scale. By focusing on maintaining desired states, the system proactively guided itself back to optimal conditions, conserving valuable man-hours.

Solution overview

Cognitive Systems’ implementation

A diagram of Cognitive System's AWS cloud computing system

Figure 1: High-level solution architecture

The previous figure illustrates Cognitive Systems’ high-level solution architecture. Data from the Internet of Things (IoT) nodes is routed to the nearest region using the Amazon Elastic Load Balancing Network Load Balancer, while client application requests are routed to Kubernetes services exposed through the Application Load Balancer (ALB). The routing of traffic to the nearest AWS Region is typically managed in a customer-specific manner, with a common practice being the utilization of Amazon Route 53. While customer-specific configurations vary, Amazon Route 53 often serves as the foundational layer. Additionally, Amazon Route 53 is instrumental in implementing High Availability (HA) within a specific region. To establish a robust, secure, and reliable communication channel between Cognitive’s IoT fleet and the Cloud, they employ MQTTS (MQTT + mutual TLS) facilitated by AWS Network Load Balancers. Additionally, AWS Lambda functions are employed for executing Extract, Transform, and Load (ETL) operations on collected data, which is stored on Amazon Simple Storage Service (Amazon S3). Amazon Athena and Metabase helped Cognitive gain and visualize insights from the transformed data. The same architecture is replicated across multiple regions to provide low latency and meet quality standards.

Automation scripts and Infrastructure as Code (IaC) pipelines have been developed, facilitating the replication of this architecture across various clusters and global regions with ease.

Conclusion

In this blog post, Mike Ravkine discussed how Cognitive Systems Corp. leveraged Amazon EKS to bring their innovative Wi-Fi Sensing technology, WiFi Motion, to reality. It outlined the evolution of Cognitive’s technology, challenges faced in latency when expanding globally, and how Amazon EKS addressed these issues. The post emphasized the benefits of using AWS for scalability, efficient management, and global reach, ultimately enabling Cognitive Systems to revolutionize Wi-Fi innovation and make motion sensing accessible worldwide.

Adopting a standard, like Amazon EKS and other services, allowed Cognitive Systems to seamlessly integrate with large customers already on the AWS cloud, significantly reducing the once-present barriers of networking, authentication, and authorization. The establishment of a secure, private link between Cognitive and its customers via AWS effectively manages typical security requirements, rendering compliance and privacy considerations more manageable and cost-effective. Cognitive conducts prototypes and trials using the public internet due to its expediency and accessibility. However, when transitioning to production, they use shared Amazon Virtual Private Clouds (Amazon VPC) and PrivateLink, as these options offer enhanced security measures and cost-effectiveness.

The partnership between Cognitive Systems and AWS, with a robust, globally accessible infrastructure, facilitated efficient scalability, intricate problem-solving, and the delivery of exceptional services. As Cognitive’s global reach expands and its customer relationships deepen, AWS and Amazon EKS continues to empower them to achieve more with fewer resources while upholding unparalleled quality and reliability.

To learn more about Cognitve Systems and WiFi Motion, visit www.cognitivesystems.com.

"Headshot

Mike Ravkine, Sr. Director (Cloud/DevOps) at Cognitive Systems

Mikhail (Mike) Ravkine is a seasoned software engineer with 20+ years of experience in the architecture and implementation of small to large-scale software systems in C/C++, Python, Perl, Java, and JavaScript (ES6, NodeJS). His expertise in cloud-native architectures, demonstrated through five years of experience with DevOps/GitOps on multiple platforms, reflects his adaptability to evolving industry trends. Throughout his career, Mike has refined his expertise in maintaining and supporting in-house verification libraries. He has overseen key aspects, including verification infrastructure, methodology, automation, and reporting, to ensure the integrity and reliability of advanced multi-core RF/DSP SoCs. As a founding employee of Cognitive Systems Corp. and now serving as Senior Director (Cloud/DevOps), Mikhail has been essential in shaping the company’s innovative solutions in Cloud-native architectures. With six years at the forefront, he leads the design, implementation, and operation of end-to-end IoT device agents and cloud backends and was instrumental in scaling WiFi Motion from a single-person initiative to a global endeavor involving multiple teams supporting hundreds of servers.

Ashok Srirama

Ashok Srirama

Ashok is a Sr. Containers Specialist Solutions Architect at Amazon Web Services, based in Washington Crossing, PA. He specializes in serverless applications, containers, and architecting distributed systems. When he’s not spending time with his family, he enjoys watching cricket, and driving.

Sukirti Gupta

Sukirti Gupta

Sukirti is Business Development Lead for AWS Startup and SMB Customers in North America, and love to help these customers on their Application Modernization journey with Container Services (Amazon ECS/Amazon EKS), Serverless Technologies and Microservices. Besides work, he enjoys spending time with his family, reading, hiking and mentoring start-ups.