ISC21 Keynote: HPC-Powered Radar Tomography for Sustainable, Equitable Cities

By Oliver Peckham

June 29, 2021

ISC High Performance 2021 kicked off yesterday with a keynote from Dr. Xiaoxiang Zhu, a professor of data science and Earth observation at the Technical University of Munich. The conference, held virtually for the second time due to the ongoing coronavirus pandemic, featured a surprisingly COVID-light agenda, seeming to focus instead on building a better new normal than the status quo that the world left behind in 2020. To that end, Zhu’s keynote spotlighted how HPC-powered Earth observation data processing, carefully applied, has an incredible amount of transformative power.

Dr. Xiaoxiang Zhu

Zhu’s presentation kicked off with a graphic of a slowly spinning Earth with a lone satellite in polar orbit around it. As the Earth rotated, that polar orbit slowly gave the satellite coverage over the planet.

“We are living in a golden age of Earth observation,” Zhu said. The game-changer, she explained, was the European Space Agency’s Copernicus program and its Sentinel satellite fleet – an “evolutionary change” in Earth observation that occurred around 2014 to 2015, resulting in an explosion in satellite data. “The data we have so far [is] around 40 petabytes, and in 2030, it will become 140 petabytes,” Zhu said. “And what is more important is that most of the data is provided by the Sentinel satellite fleet, [which is] following the so-called ‘free and open policy.’ This means these petabytes of data are accessible to everybody – and also, you, who are in the audience.”

“Without any doubt,” she stated, “Earth observation is now a big data problem.”


“Without any doubt, Earth observation is now a big data problem.”


This big data problem has big implications for a wide range of fields, from meteorology (“the classic field where we are providing information”) to “very practical things related to safety, security, mobility, resource management and … city planning.”

“In my lab,” Zhu continued, “our vision is that we want to develop innovative AI and data science solutions in Earth observation for tackling societal grand challenges – for example, urbanization, climate change, or the UN’s SDGs [Sustainable Development Goals].”

Leveraging the petabytes of Earth observation data in service of these goals means enlisting “a wide range of expertise”: exploratory signal processing techniques, data fusion strategies and much more. And, of course: “If we talk about global applications then very easily we’re ending up with petabytes of geodata, satellite, social media data. Therefore, big data analytics and high-performance computing are very crucial components for the whole harvesting process.”

One of the most impressive efforts at Zhu’s lab is a deceptively simple pitch: global urban mapping.

A disparity in geoinformation

“We are now living on an urban planet,” Zhu said. “If you look at the urban population and the rural population – the evolution over time – you will figure out in 2008, there is a secret change [in] history: there are more people living in urban areas than rural areas. And this trend is going to increase in the future.”

As an isolated datapoint, this might not seem troubling. But the geospatial distribution of that accelerating urbanization, Zhu explained, was unequal: it was mostly occurring in developing areas that lacked sufficient urban planning to ensure that the urbanized areas were also safely habitable. By way of example: “In the last decade, nearly 50,000 fires happened in Mumbai,” Zhu said. “What was really shocking is [that] nearly 70 percent were simply caused by faulty wiring.”


“We are now living on an urban planet.”


In order to address those kinds of problems through improved urban planning, though, you need to know some basic geoinformation about the region. Where are the slums? Where are people living? How tall are the buildings? What functions do the buildings serve?

“We don’t have global information in fine detail available,” Zhu said. She pointed to OpenStreetMap, which she said many Europeans might view as a substantial source of such information. But while that might work for much of Europe, Zhu explained how small OpenStreetMap’s scope was in the scheme of the world. There are roughly three billion buildings in the world. 457 million of them – just 16 percent – are in OpenStreetMap. Of those, only three percent have height information – just half of one percent of the world’s buildings overall.

“And we,” Zhu said, “are using AI and data science to close this gap.”

Closing the building footprint gap

First, the team set out to collect building footprints. They used satellite data from Planet, a new aerospace company that provided data from around 140 very small satellites that Zhu said were able to provide data with an accuracy of four to five meters on a daily basis. Using training data from 74 cities, the team taught a graph convolutional neural network to understand the relationship between the satellite data and GIS-derived building footprints.

“All together, we processed more than 200,000 planet satellite images,” Zhu said. They were able to derive global building footprints from their model, allowing them to close the 84 percent gap of buildings that did not have footprints in OpenStreetMap.

In yellow, the building footprint data that was previously available in Marrakesh. In purple, the new building footprint information from Zhu’s team. Image courtesy of Xiaoxiang Zhu.

Then came the hard part: transforming 2D satellite data into 3D data that could be used to derive building heights, as well.

Radar tomography

For that, Zhu and her team turned to data from TerraSAR-X, a German radar satellite. Here, they used a technique called “radar tomography,” which is akin to the computed tomography (CT) scans that use rotating blasts of x-rays to generate 3D images out of a stack of 2D images taken from different angles.

“Similar to CT, you are doing a scan of the patient, and each of these satellite images could be considered a tomographic slide,” Zhu said. “And every 11 days the satellite will fly over the same area, taking the image from slightly different positions. After we build up enough … images, we’re able to basically do the reconstruction with tomographic methods.”

A simplified representation of the radio tomography process (with a CT machine in the corner). Image courtesy of Xiaoxiang Zhu.

“Similar to CT, you are doing a scan of the patient[.]”


This radar tomography work is extraordinarily compute-intensive, and it is, accordingly, powered by the HPC resources from the Leibniz Supercomputing Centre (LRZ). Zhu is a self-described “power user of SuperMUC[-NG],” the center’s largest supercomputer. SuperMUC-NG, which launched in 2018, consists of 6,336 Intel Xeon Skylake-powered “thin” compute nodes (each with 96 GB of memory) and 144 Skylake-powered “fat” compute nodes (each with 768 GB of memory). All told, SuperMUC-NG delivers 19.5 Linpack petaflops, placing it 17th on the most recent Top500 list (which launched the same day as the keynote).

“We are … using a lot of computing resources from the Leibniz Supercomputing Center, because eventually if we boil everything down to formulas we are solving for each of … approximately ten thousand times five thousand pixels,” Zhu said. “For each of those pixels, we are solving a long convex optimization problem with a matrix dimension of 100 times one million, and here we get a lot of support from Leibniz Supercomputing Center and have been granted … 46 million CPU hours for our social good project.”

And now, Zhu said, “I’m happy to show you around Berlin!”

A pointillist metropolis — and eventually, a world

The detail on display is undeniably extraordinary: from the satellite data alone, radar points highlight individual statues and lampposts. The researchers could even see how buildings expanded and contracted in response to seasonal weather changes. “We are getting, in the end, a point cloud with a density of one million points per square kilometer,” Zhu said.

Berlin, as captured by radar tomography. Image courtesy of Xiaoxiang Zhu.

Combining this point cloud with the building footprints, they got building maps, complete with elevation by building. To start, the researchers worked with major cities, which offered points of comparison for validating their work. After comparing their results for 30,000 buildings with precise lidar data for the same buildings, the researchers found the height estimation to be “better than two meters.”

“For Europe, this is probably nothing really breathtaking,” Zhu said – but for areas without even building footprints available, it was something else entirely.

As of the keynote, Zhu said that the researchers had already processed more than 1,700 cities, including all cities with populations over 300,000. To complete the radar tomography globally, she said, “we can use the same technique, but we need ten times the data.”

Throughout the talk, Zhu referred back to the UN’s Sustainable Development Goals, a set of 17 objectives set forth in 2015 that range from ending hunger to ensuring affordable and clean energy. This work, for Zhu, falls under the first goal (“no poverty”) and the eleventh (“sustainable cities and communities”).

“We are making this data open and free,” she said. “And we are hoping, by providing this information, we could contribute[.]”

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

ASC24 Student Cluster Competition: Who Won and Why?

June 18, 2024

As is our tradition, we’re going to take a detailed look back at the recently concluded the ASC24 Student Cluster Competition (Asia Supercomputer Community) to see not only who won the various awards, but to figure out Read more…

Qubits 2024: D-Wave’s Steady March to Quantum Success

June 18, 2024

In his opening keynote at D-Wave’s annual Qubits 2024 user meeting, being held in Boston, yesterday and today, CEO Alan Baratz again made the compelling pitch that D-Wave’s brand of analog quantum computing (quantum Read more…

Apple Using Google Cloud Infrastructure to Train and Serve AI

June 18, 2024

Apple has built a new AI infrastructure to deliver AI features introduced in its devices and is utilizing resources available in Google's cloud infrastructure.  Apple's new AI backend includes: A homegrown foun Read more…

Argonne’s Rick Stevens on Energy, AI, and a New Kind of Science

June 17, 2024

The world is currently experiencing two of the largest societal upheavals since the beginning of the Industrial Revolution. One is the rapid improvement and implementation of artificial intelligence (AI) tools, while the Read more…

Under The Wire: Nearly HPC News (June 13, 2024)

June 13, 2024

As managing editor of the major global HPC news source, the term "news fire hose" is often mentioned. The analogy is quite correct. In any given week, there are many interesting stories, and only a few ever become headli Read more…

Quantum Tech Sector Hiring Stays Soft

June 13, 2024

New job announcements in the quantum tech sector declined again last month, according to an Quantum Economic Development Consortium (QED-C) report issued last week. “Globally, the number of new, public postings for Qu Read more…

Qubits 2024: D-Wave’s Steady March to Quantum Success

June 18, 2024

In his opening keynote at D-Wave’s annual Qubits 2024 user meeting, being held in Boston, yesterday and today, CEO Alan Baratz again made the compelling pitch Read more…

Shutterstock_666139696

Argonne’s Rick Stevens on Energy, AI, and a New Kind of Science

June 17, 2024

The world is currently experiencing two of the largest societal upheavals since the beginning of the Industrial Revolution. One is the rapid improvement and imp Read more…

Under The Wire: Nearly HPC News (June 13, 2024)

June 13, 2024

As managing editor of the major global HPC news source, the term "news fire hose" is often mentioned. The analogy is quite correct. In any given week, there are Read more…

Labs Keep Supercomputers Alive for Ten Years as Vendors Pull Support Early

June 12, 2024

Laboratories are running supercomputers for much longer, beyond the typical lifespan, as vendors prematurely deprecate the hardware and stop providing support. Read more…

MLPerf Training 4.0 – Nvidia Still King; Power and LLM Fine Tuning Added

June 12, 2024

There are really two stories packaged in the most recent MLPerf  Training 4.0 results, released today. The first, of course, is the results. Nvidia (currently Read more…

Highlights from GlobusWorld 2024: The Conference for Reimagining Research IT

June 11, 2024

The Globus user conference, now in its 22nd year, brought together over 180 researchers, system administrators, developers, and IT leaders from 55 top research Read more…

Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study

June 10, 2024

Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…

ASC24 Expert Perspective: Dongarra, Hoefler, Yong Lin

June 7, 2024

One of the great things about being at an ASC (Asia Supercomputer Community) cluster competition is getting the chance to interview various industry experts and Read more…

Atos Outlines Plans to Get Acquired, and a Path Forward

May 21, 2024

Atos – via its subsidiary Eviden – is the second major supercomputer maker outside of HPE, while others have largely dropped out. The lack of integrators and Atos' financial turmoil have the HPC market worried. If Atos goes under, HPE will be the only major option for building large-scale systems. Read more…

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Everyone Except Nvidia Forms Ultra Accelerator Link (UALink) Consortium

May 30, 2024

Consider the GPU. An island of SIMD greatness that makes light work of matrix math. Originally designed to rapidly paint dots on a computer monitor, it was then Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Some Reasons Why Aurora Didn’t Take First Place in the Top500 List

May 15, 2024

The makers of the Aurora supercomputer, which is housed at the Argonne National Laboratory, gave some reasons why the system didn't make the top spot on the Top Read more…

Leading Solution Providers

Contributors

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

The NASA Black Hole Plunge

May 7, 2024

We have all thought about it. No one has done it, but now, thanks to HPC, we see what it looks like. Hold on to your feet because NASA has released videos of wh Read more…

Google Announces Sixth-generation AI Chip, a TPU Called Trillium

May 17, 2024

On Tuesday May 14th, Google announced its sixth-generation TPU (tensor processing unit) called Trillium.  The chip, essentially a TPU v6, is the company's l Read more…

Intel’s Next-gen Falcon Shores Coming Out in Late 2025 

April 30, 2024

It's a long wait for customers hanging on for Intel's next-generation GPU, Falcon Shores, which will be released in late 2025.  "Then we have a rich, a very Read more…

GenAI Having Major Impact on Data Culture, Survey Says

February 21, 2024

While 2023 was the year of GenAI, the adoption rates for GenAI did not match expectations. Most organizations are continuing to invest in GenAI but are yet to Read more…

Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study

June 10, 2024

Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst fir Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

A Big Memory Nvidia GH200 Next to Your Desk: Closer Than You Think

February 22, 2024

Students of the microprocessor may recall that the original 8086/8088 processors did not have floating point units. The motherboard often had an extra socket fo Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire