by Matt Johannessen
Recently, I had the opportunity to lead a Novetta team tasked with demonstrating how C4ISR, IoT, Machine Learning, and edge computing technologies can improve situational awareness for disaster response operations.
Through our partnership, Amazon Web Services invited Novetta to participate in an event called Operation Convergent Response (OCR). OCR is an event hosted annually by Verizon & Nokia at the Guardian Centers facility in Perry, GA. The primary objective of OCR is to bring solution providers and first responder organizations together to demonstrate how emerging technologies can save lives. Unlike traditional technology trade shows, OCR is centered around technical solutions deployed in the field to simulate disaster scenarios. Novetta participated mostly in the “Cyber Attack on Critical Infrastructure” scenario, which demonstrated the effects of a cyber terrorist taking control of a dam and consequently flooding a small town. It’s not every day that a team of software developers gets to go into the field to participate in something like OCR. Based on our participation at OCR, AWS invited us to present at AWS re:Invent 2019 in Las Vegas during the Snow Family update.
During disaster response, knowing the location of personnel, vehicles, and equipment is critical to maintaining high safety standards and maximizing the effectiveness of first responder efforts. Tracking the location of these resources across a diverse response with multiple organizations can be challenging. Responding organizations often lack interoperability with Local, State, and Federal command centers, as well as with each other. Additionally, most available location tracking solutions depend on communication backbones such as cellular or satellite networks, which can be unreliable, cost prohibitive, and difficult to use during disaster response operations.
Our team’s solution was to deploy GPS sensors that communicate over an emerging digital RF technology called LoRaWAN which streamed location information from each sensor into Ageon ISR, Novetta’s C4ISR platform running on an AWS Snowball Edge device.
The overall architecture for our solution is to display and map track information using commodity GPS sensors dynamically assigned to people, vehicles, or equipment. The LoRa protocol supports low power, long range encrypted communications between each sensor and a central LoRa Gateway (Receiver & Packet Forwarder). The rugged AWS Snowball edge (compute optimized) with pre-configured compute instances is our processing engine and application server.
We supported local, “in the field” users as well as remote users connected via satellite backhaul viewing Ageon ISR hosted on AWS EC2. During disaster response, operations connectivity is never guaranteed, making the ability to operate at the edge without internet access critical. When internet backhaul is re-established, we can sync data collected at the edge to an instance of Ageon ISR running in the cloud. We post-processed video surveillance feeds to make use of reduced bandwidth backhaul, a typical condition during response operations.For our scenario, Ageon ISR was used to conduct safety briefings, at the edge, before each run of the scenario. Working with Verizon and AWS, we set up our solution in a mobile command center, emulating actual disaster response efforts.
Using standard Ageon ISR features, we defined hazardous areas and trigger actions such as automated camera control (a.k.a. “Slew to Queue”) and alert notifications when a sensor entered these areas. We developed a custom mobile application, API-hosted on the Snowball edge, to track sensor health, battery status, and assignment.
Using the app, we could quickly re-assign a sensor to another person, vehicle, or piece of equipment by scanning a QR code. The API fused information from the LoRaWAN gateway (sensor location, id, and battery voltage) with assignments before sending NATO standard (4676) track messages to Ageon ISR. We tuned the sensor transmit rate to ping each sensor every second while the sensor was moving. Three AAA batteries lasted on average 48 hours, better than expected given packet transmission frequency.
The LoRaWAN sensors lived up to their reputation for being “long range, low power”. We observed packets from well over one mile away using a single gateway antenna (CISCO IXM). We explored ways to tune performance with machine learning by recommending gateway placement and labeling areas on the Ageon ISR map (e.g. “popular route”, “parking area”, etc) based on conditions observed in the field. For our initial deployment, range performance exceeded expectations as evident by recorded observations from over one mile away from the gateway.
Feedback & Results
OCR was a great opportunity to demonstrate how Ageon ISR can be used to fuse emerging IoT sensor data and video surveillance feeds to produce a real time common operating picture during disaster response. By the end of the event, we had deployed sensors to 43 different vehicles, people, and boats, collected 150K+ track data points, and acquired 500 GB of video over three days using the AWS Snowball Edge (compute optimized).
Our ability to launch the entire stack within 10 minutes and re-assign GPS sensors on the fly made our solution optimal for these simulated disaster response operations. We are looking forward to additional field testing of these capabilities based on conversations that started at OCR. Special thanks to both AWS and Verizon for having us; we are looking forward to tackling next year’s challenge!