Adapting to the changing dynamics of Earth observation
Enterprises can leverage quality data using space sensing services to build innovative applications.
Space is a subject that intrigues everyone. Artificial satellites placed in the Earth’s orbit are not only used for space exploration but also for communication, observation, navigation, and broadcasting.
The evolution of satellite systems in the form of clusters has enhanced the capability and quality of Earth observation. A cluster of small satellites can revisit a specific point on the Earth multiple times a day to gather higher spatial and temporal resolution images. Compared to the traditional large satellites, these clusters obtain data at frequent intervals. This helps to monitor troop and maritime movements, detect and warn against toxic leaks in chemical plants, locate forest fires and damages caused by them, alert about oil spills, as well as provide weather updates.
However, managing a satellite constellation is not easy—it involves mission planning, satellite coordination, manual operations (that come with a human cost and effort), and the need to deliver data to an Earth station in a timely manner. Executing these tasks simultaneously is even trickier for applications that require real-time or near-real-time information, like disaster management.
To address these challenges, TCS Research recommends a satellite-sensing architecture that offers space-sensing-as-a-service. Such a platform would facilitate hosting business applications that require satellite-sensing data, enabling on-demand data gathering, onboard space computation, and timely information sharing.
Why an intelligent satellite-sensing infrastructure matters
Space-sensing infrastructure must be flexible, scalable, and smart.
Satellites currently transmit a high amount of data to Earth stations, processing it via cloud computing services. These systems have limitations, such as data storage, latency, and the lack of high-computing capabilities to process information in real time.
A satellite’s system architecture, therefore, must be flexible and intelligent to support high-end computational technology for onboard data analytics. It should also only activate communication channels to ground stations when required. It must also be scalable to accommodate high-volume data when constellations expand to integrate a growing user base. And it should be agile to ingest and process real-time data.
Orbital edge computing is one such emerging technology that can be integrated to support onboard computation, wherein a constellation analyzes sensor data in space using machine learning (ML) capabilities. It sends only quality data to Earth, making the process energy-, time-, and cost-efficient.
Moreover, satellite infrastructures should be modular so newer interfaces and protocols for command and communication can be adopted, as well as the ability to install hardware developments, such as neuromorphic chips to run ML models.
The need for space-sensing architecture.
Space-sensing architecture needs to be designed to offer modern resources, technology, and infrastructure to capture and access remote sensing data.
Generally, accessing sensor data and computations are managed as two discrete services due to the separation of sensing equipment in space and computer infrastructure on the ground. Data from multiple sensor sources are required to perform advanced data analytics. Collecting information from varied sources may cause latency in processing quality outcomes when using data to address real-time challenges.
A space sensing-as-a-service platform could synergize data acquisition, communications, and computational services to facilitate real-time satellite information.
Let’s look at its functions and features:
The platform could support dual objectives/functionalities, empowering application developers as well as users who consume sensor data. A programming framework can help design software by selecting services for data ingestion and computation, uniting them via an application. And a hosting platform could help end users request data through queries based on specific inputs. Such an application could then generate data requests that analyse information and process the final output.
The space sensing-as-a-service architecture
A service bus is a layer that helps establish a network between the user application or developer and satellite infrastructure components. The layer functions as an arbitrator, conveying user demands to resources—image or thermal sensors, computation machines, and networking channels—available in the infrastructure's back end. Uniform interfaces can be used to access system resources that call for services through virtual or digitalized layers. In short, the service bus moderates, controls, and optimizes access to resources.
To make the application perform well, a software stack is used to facilitate the three services, supported by underlying infrastructure components: Satellite-borne sensors that serve as the system’s data sources; cloud-based computational and storage components organized as cloud services; and a communication system to power connection and control between the first two.
Data acquisition service: This service serves data requests for data that is live or stored. The component can be equipped with a virtual sensor—an abstract entity that dynamically gathers data from different sensing devices based on inputs from the user application. It can combine various images acquired from a constellation of satellites, stitch them together, and present them in a single image. It can also request and deliver a data source to reduce the image resolution if the application so requires. Virtualization allows the reuse of sensor data for different purposes based on the types of requests from users, or it can incorporate a new sensing device into the system, bolstering flexibility, agility, and scalability aspects.
Computational service: Computational resources of satellite systems are usually operated from the ground. In contrast, the infrastructure component here supports a virtual machine providing cloud computing services. This virtualization powers the system with real-time computing features. Likewise, it also supports machine learning capabilities to execute analytical workflows on-board via the computing infrastructure.
Communication services: Satellite communications have several interfaces or communication channels (such as radio or microwave) through which information is passed. Software-defined networking (SDN) is a virtualization feature that accommodates multiple communication interfaces. SDN here functions as a virtual entity that ensures the desired message is transmitted to its destination irrespective of the communication medium.
Futuristic in purpose
The proposed mechanism will help organizations get real or near-real-time satellite-sensing data.
Currently, there is no comprehensive service or application that provides remote sensing information that is required by industries in real-time. The mechanism proposed is therefore futuristic in its design and purpose.
Typically, requests are sent to space agencies to collect or share downloaded data housed in their data centers. This routine can affect the research or purpose for which the information is sought, as the data ends up being dated by the time it is analyzed. Also, relying only on ground-based sensing systems is inadequate to study topographical changes, natural disasters, climate change, and monitoring landscapes. Therefore, the proposed architecture is invaluable in terms of assimilation and access to real-time satellite data.
The authors would like to thank Narayan Subramanian, former Technology head, Satellite remote sensing, TCS; Sai Prasad Parameswaran, chief technology officer, IoT and digital engineering, TCS; Dr P Balamuralidhar, distinguished chief scientist (retired), TCS Research; and Gaurav Khemchandani, former innovation evangelist, TCS Research, for their guidance and support.