AWS and ACT Escalate Video and Spatial Streaming for Emergencies

AWS ESA helicopter2

Emergency Services Agency (ESA) is the ACT (Australian Capital Territory) Government organisation responsible for emergency management services in the Canberra community and works closely with its counterparts in the states across the country.

ESA operates specialist intelligence gathering (SIG) that streams live video and collects spatial data from bushfires. Using standard and infrared cameras, SIG supplies fire line and hot spot information directly to the ESA Incident Management Teams in real time.  

Volatile Environment

Streaming live video and data from a helicopter attending a fire involves environmental challenges including limited bandwidth and constantly changing geography. Fast, efficient processing of information is necessary to achieve a reliable stream. The systems can be expensive and complex to operate, and often need costly infrastructure. Through the National Aerial Firefighting Centre (NAFC), ESA contracted helicopters and defined the requirements for a video streaming and spatial data processing system that can adapt with the changing fire seasons.

At the community level, the system had to meet the needs of emergency services at work in the ACT community and also be interoperable with jurisdictions in the other states if required. The fire season is typically four months of the year and is now growing longer, so the system needed to adapt to the peaks and troughs, achieving the best value for money over the year.

AWS ESA fireline

ESA staff at the fire line.

Scaling Up in the Cloud

ESA chose AWS because it makes a set of services available that, for their purposes, is complete and also allows them to expand according to business demands, rather than accommodating the hardware or software itself. AWS Elemental MediaLive carries out the live video processing and creates video streams for delivery to internet-connected devices. It allocates resources and manages the scaling, failover, monitoring and reporting needed to run a live video stream, in effect setting up a simple live channel very quickly.

With MediaLive in place for video transcoding and automatic archiving, the system processes incoming video to extract, process and store spatial data using AWS Lambda to run the code for these tasks. Lambda runs users’ code for applications or backend services on servers provisioned and managed by AWS. This means only the compute time consumed is paid for, while Lamda looks after tasks associated with running and scaling the code with high availability.

Automating Services

The user uploads the code and sets it up to automatically trigger from other AWS services like S3 and CloudWatch, or respond to calls directly from web or mobile apps. It could also be orchestrated into workflows. This functionality allows you to build a variety of real-time serverless data processing systems.

All of the video files and spatial data resulting from MediaLive processing are then made accessible through a secure private Amazon Simple Storage Service (Amazon S3) buckets and applications residing in Amazon Elastic Compute Cloud (Amazon EC2). Whenever ESA has to add emergency helicopters, the system can rapidly scale to accommodate the changes.

AWS ESA communitytraining

Community Training for ESA firefighters.

About the auto archiving, one of the steps in setting up a MediaLive channel is creating and automating output groups – including an archival output. ESA can review the archival video footage on demand, give access to help real-time firefighting operations and, later on, review and learn from what has happened to support future operations.

External Expertise

Bigmate, an IoT and computer visualisation and analytics company in Queensland, was chosen to help deliver the system on the ground. It oversees connected fleets, and fixed and mobile assets, through location tracking. The company understood not only the required AWS services, but also the ongoing business requirements for a system that would work and be supported continuously through the fire season, where traditional approaches would not be cost effective.

Because the levels of reliability and performance required in an environment as challenging as a helicopter attending a bushire were very high, Bigmate recognised the advantages of serverless operation – that is, dynamic management and allocation of machine resources by a cloud provider. This approach can scale and reduce the overall operational footprint needed to maintain the system.

System in Action

The system is based on MediaLive. Responsible for receiving the real-time video from helicopters, MediaLive performs the heavy processing of transcoding video, archiving it to Amazon S3, and distributing the transcoded video. Through secure web and mobile applications, end users can view live and saved video anywhere there is an internet connection.

AWS ESA Drawings2

MediaLive was attractive to ESA because it can be turned on and off as required and is highly configurable, supporting both the complex emergency environment as well as future operations, such as inserting images to screen when the video stream stops.

Lambda is employed and customised using layers that include Ffmpeg, an open source framework able to decode, encode, transcode, mux, demux, stream, filter and play most formats, up to the most recent. These are triggered from Amazon S3 to analyse and extract metadata from the video. This metadata is passed through AWS IoT device data management services, to drive the web platform.

Controlling Information

Telemetry and GPS data, collected and automatically transmitted from remote points, is embedded in the video and carries an enormous wealth of information that can be used to identify location, time, direction and other information directly from the camera and helicopter. To match this, Lambda also receives, processes and formats spatial data recorded in the helicopter by its operators, so that the information collected on fires can be continuously processed and made available to the relevant users.

AWS cloudwatch rds dash 2

AWS Cloudwatch

Whether images, shape files or KML files holding notation expressing geographic annotation and visualization for mapping and 3D earth browsers, this data can be further processed using automated instead of manual tasks when required in the future.

ESA can also use AWS supporting services such as Amazon Cloudwatch, in which activities are centrally logged and managed, and users have a unified view of operational health. It supplies data and actionable insights – logs, metrics and events – to use when monitoring their applications, responding to system-wide performance changes and optimising resource