The story of unlicensed cellular reads rather like that of mobile broadcast – if you at first you don’t succeed, try, try and try again.
The tale dates back almost a decade to the development of LTE-Unlicensed (LTE-U) for the existing 3GPP Releases of the time (10/11/12). This quickly fell foul of the WiFi community because it was designated for rapid launch in countries, chiefly the USA and China, that did not mandate use of the listen-before-talk (LBT) technique to mitigate clashes. LTE-U allowed MNOs to boost capacity without having to offload onto a separate WiFi network. It retained LTE as an anchor for signalling and control, but all data including phone calls was rooted over the unlicensed 5 GHz band, instead of the carrier’s licensed spectrum.
This set the precedent for evolving cellular unlicensed technologies, even though it failed to gain much traction, largely because performance was not compelling, the business case did not stack up, and for most telcos there was no urgent need to offload data in any case. Most had plenty of midband spectrum that was not in heavy use.
This last point also dogged LTE-U’s two successors, LTE Licensed Assisted Access (LAA) and MulteFire, which also reached a market with little demand for offload, combined with a lack of emerging new service providers that might have been eager to kick in with services over licensed spectrum. The principle was the same, to aggregate LTE in unlicensed 5 GHz spectrum with the licensed band.
LAA, introduced with 3GPP Release 13 early in 2015 as part of LTE-Advanced Pro, was little more than a standardization of LTE-U, and one that mandated LBT to support more rigorous collision avoidance mechanisms. LAA presented a common version of unlicensed LTE for the whole world, but still failed to generate much groundswell.
One reason was that it still required hardware changes to implement. This led to development of LTE-WLAN Aggregation (LWA), which required just network configuration changes because it enabled coexistence between LTE and WiFi signals. This was better than other candidate protocols for coexistence such as Light Weight IP (LWIP) because it could split individual data flows so that they could use LTE and WiFi simultaneously. As a result, applications and services could exploit the combined capacity more efficiently. So apart from reducing network deployment costs, it made efficient operation simpler, while improving performance for users.
This still did not really take off for the other reasons, and the same applied to the additional option that emerged in early 2016, MulteFire, which was more radical because it was the first to dispense with the need for an LTE anchor. This meant it could operate entirely standalone in unlicensed spectrum and laid the ground for similar capability under 5G with NR-U. But it stuck to LBT so in that sense it was a direct successor of LAA.
The objective was to combine the performance and range of LTE with the ease of WiFi deployment. Of course, some operators were hoping it would displace WiFi as customers installed MulteFire access points instead of WiFi gateways. While that did not happen to a significant extent, the combined assault of unlicensed LTE was sufficient to spook Claus Hetting, chairman of the news and advisory group WiFi NOW, into insisting that unlicensed LTE had failed largely for structural financial reasons, namely that the cost of the required densification would continue losing money for operators, while WiFi 6 would wipe out any technical lead 5G might have gained.
Both assertions are broadly incorrect. It is clear that WiFi 6 does not entirely close the gap on 5G over security and latency. And even if operators do face major investment to build out small cells, enterprises are already demonstrating that 5G alone can meet some new demanding use cases and can be cost effective, not least because there is no alternative. The extent to which operators can make indoor enterprise 5G services pay remains to be seen, but already there is mounting evidence that they will for the right use cases and sectors with the help of network slicing to emulate the security and performance of private networks.
There are two factors conspiring to make NR-U an exciting proposition for the industry as a whole, entry of new players and availability of suitable midband spectrum at 5 GHz and 6 GHz. This applies equally to the two flavors, Anchored NR-U, combining licensed 5G spectrum with unlicensed, and Unanchored NR-U, which is a successor of MulteFire, operating standalone in unlicensed spectrum only.
The cellular community, spearheaded by Qualcomm, which has almost single-handedly carried the torch for unanchored unlicensed technologies, contends that standalone NR-U has been designed from the ground up to coexist fairly with other unlicensed technologies including LTE-LAA as well as WiFi. But the WiFi camp has still not been mollified and this whole issue of coexistence is still not full resolved, as can be seen from the fact that the IEEE, as recently as January 2021, published a paper proposing a new method of handling collisions that has been shown to improve performance for both 5G and WiFi, but especially the latter.
The paper notes that NR-U currently uses a form of the traditional CSMA-CD (collision sense multiple access with collision detection) that was pioneered in the pre-mobile era for Ethernet during the early 1980s to ensure fair communications for multiple users over a shared physical communications medium.
As part of the coexistence protocol, base stations have an additional constraint in that they can only send signals periodically at fixed time intervals to avoid clashing too much with WiFi. But then in order to mitigate the negative impact of that on 5G performance, some schemes propose sending a reservation signal between the end of the CSMA/CD back-off procedure and the next base station transmission time, to prevent nearby devices from accessing the channel in the meantime. This restores some 5G performance but at the expense of WiFi, which loses out while that channel is hogged.
The IEEE paper proposes a new scheme under the heading of CR-LBT (listen before talk with collision resolution) that it claims greatly decreases channel resource waste caused by collisions and improves channel resource sharing fairness. This, the paper’s authors contend, can increase the throughput of both NR-U and WiFi networks, the latter by up to three times. While technical details are beyond our scope here, the main point is that it reduces the overhead of mutual backing off when collisions occur, while also eliminating that reservation process, freeing up WiFi data for transmission during that time.
So, while coexistence is still work in progress, at least under Release 16 NR-U is ready to roll out on unlicensed 5 GHz bands. This not only liberates new bandwidth, but more brings flexibility over how it can be exploited both indoors and outdoors. Then the 6 GHz bands offer greenfield spectrum for indoor environments with natural isolation from the outdoors given the difficulty signals have traversing walls.
As regards competition with WiFi, Hetting is both right and wrong. He is right that WiFi will continue expanding within its traditional domains under its sixth generation, but wrong to assert that 5G Unlicensed will register little more than a pin prick. The main impact of NR-U will be in new domains or use cases, although it could be argued that WiFi 6/6E was never a contender for some of those anyway, because of the range required, or need for ultra-low latency.
It is clear though the way unlicensed spectrum is being allocated leaves room for both. That is most obvious in the USA where the FCC has made 1200 MHz of spectrum in the 6 GHz band available to both WiFi and 5G NR-U.