Testbed Transports Wireless Innovation
A communication research lab provides a hub for real-life outdoor testing of advanced technologies.
Across 15 blocks in New York City sit the beginnings of an extensive wireless testbed, which will help advance driverless car, smart city and other technologies for the modern urban environment. The outdoor laboratory, known as COSMOS, provides a platform for researchers to experiment with a low-latency, ultra-high bandwidth wireless network during everyday life in West Harlem.
COSMOS, or the Cloud Enhanced Open Software Defined Mobile Wireless Testbed for City-Scale Deployment, harnesses an advanced optical network and powerful computing to support phased array antennas, software-defined radio platforms, edge cloud applications, advanced communication capabilities, industrial controls, and high-bandwidth sensors, among other technologies. Researchers from academia, industry and the government, with approval, are able to use the testbed to examine emerging wireless solutions by logging in remotely, explains Gil Zussman, professor of electrical engineering at Columbia University, who leads Columbia’s contribution to COSMOS.
“We have more cellphone users than ever on the planet. There are many more devices and applications, data rates are going up, and there is definitely a need for research in wireless from the application point of view,” says Zussman, who also heads up the Columbia wireless and mobile networking laboratory, known as WiMNet.
“From the physical layer point of view, there are many technologies that have improved or have emerged that will go into 5G or will go beyond 5G systems, including millimeter wave or full duplex, which the military calls STAR—simultaneous transmit and receive. These technologies are out there and they have been tested in a lab, or in a single type of network, but they have not been tested at scale,” he states.
COSMOS is part of the National Science Foundation’s (NSF’s) Platforms for Advanced Wireless Research (PAWR) effort, which is establishing three additional urban testbeds for wireless technology research. The NSF also created the PAWR Industry Consortium to leverage the expertise of and contributions from corporate members, including Verizon, AT&T, Sprint, T-Mobile, Samsung, Nokia Bell Labs, Oracle, Keysight Technologies and National Instruments, among many others.
“It is a unique aspect of the program that they have commitments from 30 industry consortium members to provide some kind of support to the testbed,” Zussman continues. “And they can also use the testbed for their own research. There is a lot of interest in edge cloud and in computing in general as we move to 5G and beyond.”
Through NSF funding and the consortium, Rutgers University is leading the COSMOS effort, with Columbia University and New York University (NYU) also jointly developing the wireless testbed. Other project partners include: the city of New York, Silicon Harlem, City College of New York (CCNY), University of Arizona (Arizona) and IBM. IBM is providing millimeter wave support, while Arizona provides the key optic solutions. Optical networking is intrinsic to next-generation communications, the professor says, as it provides fast connectivity between edge cloud computing and radio nodes.
“We want to have an optic connection to all of the sites,” Zussman specifies. “Since we want to provide experimenters the flexibility to create whatever backhaul and fronthaul they want to have for their experiments, you also want to provide them very high bandwidths. You don’t want the wireless network to operate at a gigabyte per second (Gbps) but then get stuck with some [slow] Ethernet link.”
Once fully deployed, the testbed infrastructure, across 1 square mile in West Harlem, will include nine large sites, which are rooftop applications of multiple phased array antennas and edge computing; 40 medium sites with street-level sensors, dual-use devices for wireless and wired backhaul capabilities, such as GPS or smaller antennas; and 200 smaller portable nodes, the professor shares.
The partners started with a pilot effort in May 2019, deploying two large, three medium and 30 small nodes and fiber connections, including an underground link provided by the city to lower Manhattan and NYU’s data center. The group set up the control center at Rutgers and at Columbia, installed an optical core and computing infrastructure and linked connections to the school’s data center. They also mounted antennas atop Columbia’s Schapiro Center for Engineering and Physical Science Research building on 120th Street, between Broadway and Amsterdam Avenue.
The link to edge cloud capabilities and the fiber connectivity provides a lot of flexibility to support a range of experiments, Zussman offers. “The fact that you have something 7 miles away from us, and you can decide if you do your processing here or there, provides all kinds of experimentation capabilities,” he notes. “The idea is that it is not just for us. Anyone can log in [who is approved] and get the testbed for a few hours and run the experiments.”
Meanwhile, phase one of the COSMOS testbed deployment will begin next September, which will include installation of two more large sites at CCNY, four medium sites at Columbia and a co-location site with NYU’s data center. Phase three of the project, which is contingent on additional funding, the professor says, would involve a much more widespread deployment of technology across light poles, public housing and public schools, in addition to six more large sites across the 15-block city footprint from 120th Street up to 135th Street in West Harlem.
The characteristics of COSMOS, with a low latency of less than 5 milliseconds and edge computing between 10 to 100 giga instructions per second (GIPS), allows researchers using the testbed to experiment with millimeter wave technologies that run at a Gbps rate. For connected cars, and in the future, driverless cars, this means the ability to leverage information from roadway sensors, lighting, images and video, cloud infrastructure or other situational awareness information from a busy city intersection.
“Latency and compute are now important metrics, especially if you want applications that operate in real time at the edge of the network, such as driverless cars and augmented reality,” Zussman states. “These technologies need very fast response, and for that you need latency of below 10 milliseconds, and you also want to place your compute power close to the edge. So, our objective is to allow researchers to experiment with ultra-high bandwidth, at Gbsp, latency below 5 milliseconds latency and edge computing, meaning we’ll have administered cloud to the edge.”
As such, the testbed infrastructure will provide a platform for researchers to tackle key next-generation wireless capabilities—optical networking, millimeter wave communications; full duplex; and massive multiple-input, multiple-output (MIMO) technologies—which are commonly seen as a paradigm shift, he notes.
Full-duplex capability, or the ability to transmit and receive radio signals on the same frequency, saves on spectrum usage because it effectively doubles the capacity by reusing the same spectrum for uplink and downlink of communications, greatly saving operator spectrum purchasing costs. Zussman and fellow researchers from Columbia are engaging in full-duplex wireless design experiments to assess the network protocols needed for full-duplex nodes.
Meanwhile, massive MIMO, with its large-scale antenna systems, is seen as a foundational element to a 5G network. It allows a large number of users to transmit and receive multiple signals at the same time on the same radio frequency band. The multiple antennas enable great improvements in throughput and radiated energy efficiency, according to IEEE researchers Erik Larsson, Ove Edfors, Fredrik Tufvesson and Thomas Marzetta, in their study, Massive MIMO for Next Generation Wireless Systems. “Other benefits of massive MIMO include extensive use of inexpensive low-power components, reduced latency, simplification of the media access control layer and robustness against intentional jamming,” the researchers indicate.
Researchers can also leverage software-defined radio technologies, software-defined networking and cloud computing. “And as a software-defined platform, we want it to be programmable,” the professor states. “We don’t want to place black boxes from certain providers. We want to provide open source and programmable infrastructure for the mobile wireless testbed at city-scale deployment.”
The professor adds that a key Federal Communications Commission (FCC) decision will help ease licensed researchers’ access to the testbed. “Several weeks ago, the FCC announced that the area is one of the two first innovation zones in the country,” Zussman emphasizes. “This means that if you have an existing FCC license to run experiments in a frequency say, in Virginia or New Jersey or wherever your lab is located, and you want to bring this experiment into the city, and you are operating in one of the frequencies in which we are allowed to operate, you don’t need to reapply to the FCC for a new experimental license. This has been streamlined. This is great for us as a testbed … as well as for experimenters.”
Another important aspect of the COSMOS testbed effort is community outreach. “It’s been very important to work with the local community here in building the testbed,” Zussman notes. “And we’ve been working with them on many initiatives for the community, and this has been orchestrated by Silicon Harlem and the Columbia government community affairs department.”
As part of COSMOS, the partners included an educational component for area elementary, middle and high school students, as well as a summer training program for teachers. With stakeholders, the partners helped develop a curriculum for the students, including related science experiments and research opportunities for laboratories that can be run remotely on the testbed.