How GeoSpock is supercharging geospatial analytics

Storing and analysing huge volumes of geospatial data from internet-of-things (IoT) devices using traditional databases can be tedious and expensive, with analytics applications often plagued by high latency and slow response times.

That was the problem that the founder of GeoSpock was trying to solve with GeoSpock DB, an analytics database optimised for high-performance querying and data fusion on real-world use cases. The company was founded by Steve Marsh in 2013 while he was pursuing a doctorate in computer science at the University of Cambridge.

At the time, he was building a supercomputer to mimic the human brain using biologically inspired, massively parallel architectures – a concept that still drives the design of GeoSpock DB.

Speaking to Computer Weekly in an interview, the company’s head of data science, Felix Sanchez-Garcia, said GeoSpock DB is aimed at optimising geospatial queries that often relate to where a device or item is at any point in time.

For example, an organisation may have a global geospatial database but is only interested in analysing a small area within a city. With GeoSpock DB, he said, the organisation can run and optimise that query using only SQL (structured query language), without the need to go through petabytes of data, making the process more efficient.

According to Sanchez-Garcia, GeoSpock DB differs from traditional databases in a few ways. First, by separating storage from compute, queries can be more efficient, especially when querying large datasets. He noted, however, that this is not unique to GeoSpock as Google’s BigQuery data warehouse has similar properties.

Another difference is in indexing. Sanchez-Garcia said that while traditional geospatial databases can only index “fairly small datasets”, GeoSpock DB can do so for much larger datasets in the range of terabytes and petabytes.

As a cloud-based database, GeoSpock DB is typically deployed in a customer’s own Amazon Web Services (AWS) cloud environment, although the company is looking at ways to bring secondary compute to the edge of a network where IoT devices may reside.

On the types of applications that might benefit from GeoSpock DB, Sanchez-Garcia cited IoT applications in the automotive industry that have amassed large volumes of data.

“Analysing all that data is very cumbersome and expensive as it requires a data scientist to interact with very specialised tools,” he said. “It’s not something you can productionise with self-service dashboards.”

Also, Sanchez-Garcia said urban planners can use GeoSpock DB to query traffic conditions within seconds, while retailers can use it to ascertain footfall before deciding if they want to open a store at a given location.

To demonstrate the potential use cases in smart city planning, GeoSpock has also developed a synthetic dataset based on Singapore’s vehicle dataset, successfully applying its database in use cases such as congestion pricing and traffic analysis.

Sanchez-Garcia declined to reveal whether GeoSpock is working with specific government agencies in Singapore, noting that the company has been eyeing innovation hotspots to test its technology ahead of the rest of the world.

Moving forward, Sanchez-Garcia said GeoSpock is looking to make GeoSpock DB available through other public cloud providers, as well as containerising the product so that it is cloud agnostic and can be deployed in on-premise environments.

In October 2020, GeoSpock raised $5.4m in additional funding, bringing its total funding to date to over $32m. The latest funding round was led by nChain, a global digital ledger research and technology company and Cambridge Innovation Capital.

Other GeoSpock investors include NTT Docomo Ventures, Global Brain, Parkwalk, KDDI Innovation Fund, 31 Ventures and Meltwind.