Radio-Interferometry
My research career began with trying to solve an open problem in designing robust high-capacity wireless networks: self-interference, an essential consequence of the broadcast properties of the wireless medium. The wireless medium is a shared resource, and thus if nearby devices (radios) transmit at the same time (and at the same or even neighboring frequencies), the transmitted signals collide, resulting in interference. In traditional network designs, interference is considered harmful, and consequently, networks have been architected around the conservative principle of avoiding interference whenever possible, which in turn limits the capacity of the wireless network.
``It is generally not possible for radios to receive and transmit on the same frequency band because of the interference that results.’’ Wireless Communication, Andrea Goldsmith, 2005 [Chapter 7]
The quote above captures a long-held assumption in wireless communication that a full-duplex radio, or a radio that can transmit and receive at the same time on the same frequency, was impossible to create. This assumption has led to the design of all networking layers for wireless communication using handicapped half-duplex radios. My research invalidates this fundamental assumption. The key challenge in full-duplex radios is that the radio’s own transmitted signal causes a strong interference to the signal it is trying to receive. Furthermore, the transmitted signal is a trillion times stronger than the noise floor of the receivers; therefore, radios need to cancel the transmitted signal to the noise floor to enable proper full-duplex radios. Even if somehow, one could achieve high precision ADC with high bandwidth, it’s impossible to cancel the transmitter noise and other non-linearities of radios. I designed and prototyped new techniques to infer and altogether cancel this self-interference, thus enabling devices that can transmit and receive simultaneously in the same frequency band. The key observation was to dissect the problem with the cancelation requirement for each underlying impairment. The dissection was just part of the solution; the masterstroke was analog-sinc interpolation filter design, which allowed the cancellation to re-create self-interference accurately and cancel it. The sinc-interpolation techniques are well known for digital signals. Their extension to analog signals was never thought of; my work builds an analog sinc interpolation technique, which enables the ability to achieve such a huge cancellation requirement. The impressive fact is that my work executed the above with RF-circuits and mixed-signal design with a complete digital signal processing to build an actual full-duplex radio.
The next fundamental challenge was the scalability of full-duplex radios, i.e., enabling MIMO full-duplex radios, which requires cancelling quadratic number of signals to be canceled (signal from each antenna to every other antenna – N^2 signals), thus a huge complexity in realization of scalable full-duplex radios. I developed techniques to reduce the cancellation complexity to a linear scaling. A surprising realization during my research was that the ability to listen while transmitting is a fundamental capability that has applications well beyond wireless communication, passive-wireless sensing of the physical world, IoT connectivity, all solving fundamental problems. For example, if a relay wants to receive a signal from a neighboring node and simultaneously transmit the same signal to another node, or if a WiFi radio wants to receive the reflections of its transmitted signal, in both cases, it faces the challenge of receiving while transmitting. The next major contribution of my research is to enhance and develop the fundamental techniques of self-interference cancellation as tools, upon which a variety of systems can be built-in wireless networks, wireless sensing, IoT connectivity, and virtual reality.
I have been working towards building an entire stack from RFIC to networking for full-duplex radios in recent times. A common misconception is that full-duplex would only double the spectral efficiency; it can re-design the networks. Achieving full-duplex radios would transform the wireless networking as we imagine; a plethora of problems in wireless communication like rate adaptation, packet collisions, or unreliable connectivity due to the varying nature of wireless channels all could be solved. Specifically, if we can listen to the receiving link in real-time on the same frequency and adapt radios transmission on the same frequency accordingly, all the above-stated problems would be solved. Specifically, we have two unpublished works demonstrating the RFIC development combined with network stack showing 10x data-rate improvement for LoRa.
The fact is that my work inspired full-duplex radios research in the integrated circuit community; most would expect such a radio advancement and development to take place in the circuits community and systems built over it; it’s impressive to see the other way around. This work has my unique signature combining techniques from signal processing, algorithms to RF circuits and hardware. My work has inspired communication theory researchers to leverage full-duplex radios to solve a variety of problems from rate-adaptation, collision avoidance and low-latency networks, and many more. On the other hand, the networking community is exploring new sensing and communication applications that can be built with full-duplex radios, as it enables radios to transmit and receive simultaneously. It started an entire area of full-duplex research and its applications to many different topics.
The full-duplex work not only impacted the academic world but also led to commercial success. Full-duplex was considered a near impossibility that ability to achieve it inspired the entire industry to support it for a commercial product; to some extent, the belief was it doesn’t work as claimed. It led to the foundation for a startup Kumu Networks, which has raised $55 million till-date. Based on my work on full-duplex radios, Kumu Networks has successfully raised extensive capital and demonstrated field-tested solutions for full-duplex radios. I led the product development of full-duplex radios at Kumu Networks, which was field-tested by Tier-1 networks. In a matter of six years, it went from impossible to feasible to an actual product!