SBIR-STTR Award

Low-power processors, intelligent wireless networks, and low-power sensors coupled with Distributed Machine Learning (ML) running at the edge of the network, enables a multitude of sensors to be put anywhere: not just where communications and power infrastructure exists, but anywhere valuable information is gleaned. Pushing Machine Learning to the edge allows for immediate availability of predictive and intelligent classification and regression abilities, which is not dependent on available bandwidth/connectivity/latency of going back to the cloud, perform ML, and send results back to the edge. This is particularly important for a network designed to operate in a remote and fragile setting subject to unforeseen interruptions. The way these sensors are networked determines whether the sensors can provide reliability in face of disruptions. The platform/network being proposed has been architected with the following criteria in mind: Reliability Security Sensors can be put anywhere Ease of deployment Modularity Cost Our proposed solution is a Mobile Adhoc Network comprised of two types of nodes: Mesh Nodes, and Cellular Nodes. All nodes are communicating with each other at the physical layer using their own built-in Wifi, and at the link layer using the B.A.T.M.A.N. (Better Approach to Mobile Adhoc Networking) advanced link layer mesh routing protocol. Link level routing (as opposed to Network level routing) uses raw ethernet packets to transfer routing data, making the network very resilient to unexpected interruptions in the network. This is because knowledge about routing is distributed throughout the network. Therefore, all nodes appear to be link local and are unaware of the network's topology as well as unaffected by any network changes. It is our intention to prove that optimization of this protocol will provide the best results for a resilient network in remote and fragile environments. The choice of processor for the Mesh/Cellular Nodes were done carefully to be able to perform all necessary processing (for example Machine Learning predictive model execution), while maintaining low power, enabling all nodes to operate via solar power. It was proven that intelligent processing on the IoT processor enables utilization of cheaper & lower power sensors. Various Machine Learning Algorithms were evaluated, and ultimately Random Forest was chosen for three reasons: Random Forest shows best performance Random Forest shows strongest robustness to loss of data Random Forest lends itself very well to a distributed implementation across many nodes We intend to show that a distributed implementation of the Random Forest algorithm at the application layer across the logistics network, coupled with the distributed routing intelligence of the BATMAN protocol, will enable the most agile and robust network for digital logistics needs, particularly in remote and fragile environments.

Benefit:
Two immediate and important applications that have been envisioned for the proposed network are early detection and growth prediction of wildfires, and low-cost fever detection network for early screening for COVID19 in public spaces. The solution architecture is highly modular, and we have demonstrated the ability to very quickly adjust the nodes for a given application. These two applications have specifically been chosen due to their potentially HUGE (multi-$Billion) market demand/size, as well as urgency to deal with. The value proposition of this solution in both use cases is very clear: Wildfires: the main problem is early detection, and existing solutions suffer from timeliness of response as they arent cheap/easy enough to be deployed at exact high-risk locations. Our solution can be deployed easily to exact locations with high risk of fire starting (next to high voltage power equipment.) Public Fever Detection Network: existing solutions are expensive/difficult to deploy. Our architecture enables order of magnitude cheaper cameras that can easily be deployed anywhere. This is particularly important for all restaurants/schools/malls. who are already financially hurting. The solution and the preliminary prototypes have been discussed and demos done to the Red Cross of Southern California, who have expressed great interest in both. The plan is once a realistic working prototype is developed (at the end of Phase 1), the Red Cross would put us in touch with local and state fire authorities to start thinking about test deployments in realistic settings (which would be Phase 2 of the project). We believe the Red Cross is an excellent channel to take both products to market very effectively.

Keywords:
IOT, IOT, Distributed Routing, Machine Learning on the Edge, sensor fusion, Disaster Area Networks, Mobile Adhoc Networking, Machine Learning, Agile Systems

DON is seeking to evolve the complex logistical support needs of a system that could be made remote and inaccessible by external events (e.g. health emergency, natural disasters, wartime). Ryalinks is a startup focused on utilizing Machine Learning towards solving highly complex verticalized problems. We work closely with UCLA and UCI and their respective Data Science and Ultra Low Power labs. Prior to Navy engagement, Ryalinks had developed a highly resilient ML driven mesh network solution designed to operate in remote and fragile environments. This platform was a good fit for the DON requirements of a robust logistical support network, operating in remote and fragile environments. As such Ryalinks was chosen as part of Phase I. During Phase I, Ryalinks expanded the platform to include Ultra Low Power sensor nodes, as well as the ability for the ML algorithms to perform Predictive Diagnostics prior to a fault arising. The Ryalinks solution can be deployed on various Navy platforms such as ships or aircraft to capture important data about the operations of various sections, and in real-time assess the operational health and predict when a problem might occur or detect the problem quickly after it arises. The Ryalinks platform can be a superior solution for the Navy because: Battery-less acquisition: Ambient energy harvesting solutions enable a broader array of sensor/measurement possibilities for the Navy: for example while a simple temperature sensor can operate with two AA batteries for years, an infra-red camera will need a harvesting solution Link-Layer Routing: Self Discovery - Self Healing - Scalability Custom Machine Learning for the Navy: Original research regarding Machine Learning on the edge, designed to operate in real-time with limited and incomplete data During the Base period of Phase II, Ryalinks will transform its platform to allow for operation in realistic scenarios and environments: Create the Ryalinks central platform Industrialize the Ryalinks Micro-Mesh Validate and thoroughly test power scenarios of the ultra-low power sensors Create and test simple Macro-Mesh structures Implement and validate end to end security Get necessary certifications Investigate 3 relevant use cases with the ML algorithms NASA TurboFan jet engine Remaining Useful Life (RUL) estimation Predictive Maintenance on a Naval Vessel Propulsion System Combined Diesel Electric & Gas (CODELAG) Combined fixed camera and drone-based monitoring of high-power transmission lines for corrosion detection. Alternatively, can be switched to monitor naval vessels using the same camera sensors/drones for areas of rust Test the Micro-Mesh and Macro-Mesh in a high-fidelity lab simulateing realistic operating conditions During the option period: Develop connectivity between the Ryalinks Platform and 3rd Party Cloud platforms including potentially the Navy back-end Work with the Navy or a Navy approved testbed provider to test the end to end solution in a naval test-bed

Benefit:
The Ryalinks platform is capable of predicting infrastructure risks/hazards either before they happen, or detect them shortly after they happen. The primary commercial market being targeted is the Energy & Utility industry (electric, oil and gas). The highest growth in IOT industry is shifting from the consumer sector to the enterprise/industrial sector, reaching 14B connections by 2025. Within the enterprise/industrial IOT sector, utilities are the largest market opportunity with over 1.3B connected sensors worldwide. Furthermore, within the enterprise/industrial use cases, specifically, Predictive Maintenance is one of the most promising applications demonstrating a 40% CAGR year over year the next 5 years, reaching a value of over $25B by 2023. One of the key capabilities of the Ryalinks platform is the level of intelligence it can provide at the edge of the network, in real-time. Edge ML allows for rapid understanding of the data being captured by the sensor network and facilitating real-time reaction. One of the key differentiators of the Ryalinks platform vs. other players in this market is that the functionality of the platform can switch in real-time between the two primary capabilities: Pre-fault: Predictive maintenance: prior to a fault, being able to predict the Remaining Useful Life (RUL) of a system or component. Post-Fault: Early Hazard Detection & Growth Prediction: once a fault or hazard occurs (fire, gas leak, water pollution), enable very quick detection before it spreads widely, and leveraging machine learning to predict how it might grow. Electric utility companies are among the most asset-intensive industries and one of the most critically essential, where failure can mean catastrophe. Historically, utility companies have utilized preventative maintenance tactics to maintain the electric grid reliability, which involves following a predetermined fixed schedule to regularly inspect and service assets. Predictive maintenance, on the other hand, is not dictated by a predetermined schedule; rather, targeted areas for maintenance are dictated based on analytics. Predictive analytics offers utilities enormous savings in maintenance expenditures, which can account for 20-30 percent of operating expenses. Additionally, almost 80% of the devastating wildfires ravaging the western US are caused by humans, and most of those accidentally by infrastructure malfunction. A specific example is the Camp Fire of November 2018 in California (deadliest fire in California history) which was ignited by a spark from high voltage equipment of PG&E that resulted in 83 fatalities and $18B of damage and pushed PG&E into bankruptcy. This suggests that electric power utility companies should be very interested in a platform that has both pre-hazard and post-hazard capabilities. Preliminary discussions with utility companies have validated this conjecture.

Keywords:
IOT, Digital Logistics, self-healing networks, Predictive maintenance, Mesh Networking, Machine Learning, Edge Computing