Kumu’s new chip sees its focus shift from full duplex to multiband WiFi

Kumu Networks is best known for its work in full duplex radio, a technique which promises to increase spectral efficiency dramatically because it allows transmit and receive to occur in the same channels. Though this is a technology that has already been implemented in some wireline networks, it is considered cutting edge in wireless, and has been heavily associated with 5G. Like many ‘5G’ innovations, however, it is likely to find its first commercial home in other sectors, such as backhaul and WiFi.

So Kumu’s first trials were in fixed wireless/backhaul, and the RF chip it has released this month for customer evaluation seems to be targeted primarily at WiFi, and particularly at the new spectrum allocations, in the 6 GHz band, which are being opened up for unlicensed use in the USA, and potentially elsewhere.

In that context, the interest Kumu hopes to arouse is not so much about its full duplex capabilities, but its related innovations in self-interference cancellation (SIC), a technique which mitigates the interference that arises when two radios in adjacent bands are included in a small unit – such as a future WiFi device supporting both 5 GHz and 6 GHz frequencies.

Kumu gained prominence with full duplex trials with some high profile names like Deutsche Telekom and NTT Docomo. But a chip technology that, however innovative, currently remains in a niche, needs to ride any wave that comes its way. So the firm is promoting the potential for its KU1500-0 SIC chip to enable service providers to make the best use of 6 GHz (5925-7125 MHz), a band which the FCC approved last month for unlicensed use.

Assuming the FCC’s proposal is ratified, after a comment process, it will be important for WiFi vendors to be able to support all three of the standard’s bands, 2.4 GHz, 5GHz and 6 GHz, in a single unit. That is where Kumu’s technology comes in, promising to suppress the self-interference that could occur in such devices, especially because there will be only a narrow guard band between the two higher frequencies.

That guard band is insufficient for a conventional RF filter to handle, according to Joel Brand, Kumu’s VP of product management. The company’s chip can be used to enhance the performance of those filters, or replace them completely. He told FierceWireless that access point manufacturers would have to “be very creative about how they put 5 GHz or 6 GHz in a single AP … They would have to put filters on top of filters.”

He said there are planned tests in other applications where self-interference threatens to be a big issue, such as between WiFi and IoT radios like Bluetooth in the 2.4 GHz band. As well as its original use case of full duplex spectral efficiency, the Kumu technology will also be tested with a view to reducing size, weight and isolation requirements for mobile base station filters and for the antennas which go into repeaters, relay nodes and RFID readers.

The new KU1500-0 was designed for a 2×2 MIMO radio and incorporates 300 nanoseconds of delay elements for handling interference from RF reflections. It was made using an 130nm RF SOI process, while the next version, due next year, will move to the most advanced 45nm RF SOI technology, announced earlier this year by Global Foundries.

Kumu addresses the interference issue by connecting a smart echo-canceller in place of the usual RF rx/tx combiner/splitter in a standard base station. The Kumu radio generates an additional signal that cancels out the interference – a different approach to established interference cancellation techniques, and one which could be applied to cellular networks or to WiFi.

According to Kumu, the system is constantly scanning the radio environment, gauging the state of the airwaves at any given time, as well as the way the transmission is changing the signal environment at the receiver. Its algorithms then compensate for those changes as the signals arrive from the opposite direction, in effect applying echo cancellation to radio waves.

Kumu insists it is not defocusing on full duplex, but it clearly feels SIC is a greater differentiator. There are several start-ups and academic projects focused on full duplex, as well as some commercial chip providers in the wireline world. So Kumu is one among many, and the crowd could get larger, given that both the major wireless radio standards bodies, 3GPP and IEEE, are considering the technology.

CableLabs has decided the next generation of its DOCSIS standard will support full-duplex at the cable modem termination unit, and Cisco is believed to have contributed algorithms to that effort. The 3GPP is evaluating full duplex as a possible element in Release 16, the second wave of 5G standards. It has examined simulations that showed 70% capacity increases for base stations, and Cellcom says it achieved 50% increase using Kumu’s technology in an LTE network in 1.8 GHz spectrum, relaying signals over 1,350 meters in a residential neighbourhood.

Kumu has investment from Deutsche Telekom, Swisscom, Verizon and Cisco, among others. In October 2016, DT conducted the world’s first public operator trial of self-interference cancellation, a technology which has been defined as a Technology Building Block for 5G by the NGMN Alliance.

DT said SIC technology could be used in existing networks and potentially solve small cell issues by using the same spectrum for access and backhaul. Spare macrocell capacity created via this approach could be harnessed for backhauling nearby small cells, dividing the full 100Mbps of capacity in the macro sector between several metrocells, which would not need their own backhaul, and could also act as relays. In this scenario, the small cell would need to implement the special receiver but no special handsets or macrocells would be required.

The operator’s 5G:haus research arm conducted the tests in Prague, Czech Republic, with Kumu, to determine whether SIC could enable full duplex communication, and whether that would remain robust and stable in real world conditions.

However, there are downsides to full duplex – limited coverage, and interference risks despite SIC. Researchers at New York University Tandon School of Engineering and Trinity College, Dublin last year proposed a hybrid technology which would address these issues.

Shivendra Panwar, professor of electrical and computer engineering at NYU Tandon, says full duplex networks create higher levels of interference, despite SIC and this leads to “many more outages and dropped calls”. The researchers believe the advantages of both systems could be realized with a hybrid network that mixes full and half duplex cells. Carriers could then optimize their networks to meet their particular levels of demand and spectral efficiency without excess interference.

Kumu is just one company which is leveraging interference to improve performance and spectral efficiency. In June, GenXComm raised seed funding for its own full duplex development, while MimoTech is another pioneer in this area. Magnacom proposed a new modulation technology, WAM, to improve on QAM, and was acquired by Broadcom. And Artemis emerged in 2014 with its technology to harness interference with distributed antennas to create pCells.