Gigabit LTE is becoming a reality today and being strongly hyped at the moment, combining several techniques and frequency bands to achieve headline peak data rates. We review the different alternatives and consider if this is a threat to 5G.
Exactly what is a gigabit data rate?
You may think this is straightforward, with Gigabit data service meaning that a 1 billion binary bits of data are transported from one place to another in 1 second. You'd be correct.
Those who thought it meant 230 bits (1 073 741 824 bits) per second should be calling that a Gigibit instead according to Wikipedia (slightly different spelling) and about 7% larger.
There may also be some confusion about storage space where Gigabytes are 109 versus Gigibytes 230. So with 8 bits per byte and including some signalling overhead, you’d need about 10 seconds to transfer a Gigabyte of data.
Take this figure with a pinch of salt because the data stream includes many overheads, such as TCP/IP packet headers and signalling information other than the directly useful content itself.
Downlink or uplink or both
There’s also plenty of potential for confusion here. As we explained in detail when reviewing Wi-Fi Access Point data rates, wired gigabit Ethernet cables are full duplex and can sustain 1Gbps rates simultaneously in both directions.
FDD (Full Division Duplex) LTE can also achieve high rates in both directions, but generally the uplink performs less efficiently than the downlink even although the same amount of dedicated spectrum is available.
TD-LTE shares the same spectrum band for both uplink and downlink (as does Wi-Fi), and often it’s the total data rate of both that is touted as the peak transfer rate.
In the majority of cases, data download rates are of greatest interest because most services are driven from the Cloud.
Achieving higher data rates
The three fundamental methods of increasing capacity on the wireless network are:
- Increasing the spectrum used
- More efficient used of spectrum
- Re-using the spectrum
Achieving higher data rates in a lab environment requires (1) and/or (2).
Achieving higher data rates where many users are competing for service requires (3) and a combination of either/both (1) and (2).
Using more spectrum
Operators are always looking for more spectrum to use in the busiest and dense areas. This can be allocated separately to different smartphone users within the same cell to increase capacity, avoiding the need to deploy additional cells.
Using a feature of Carrier Aggregation, multiple frequencies can be used simultaneously, streaming the data in parallel and combining it into a single stream at both ends. 3GPP standards initially supported up to four different frequencies with later standards increasing that further.
Spectrum bands can be within the regular licenced frequencies, typically 1800 to 2600MHz. Those operators with plenty of spectrum to play with can achieve a lot with this. For example, September 2016 saw EE UK combined 20MHz at 1800MHz with 35MHz at 2600MHz to achieve rates of 450Mbps using the latest Cat 9 smartphones.
Other spectrum choices include using standalone CBRS, where a demonstration by Verizon, Ericsson, Qualcomm and Federated Wireless used the maximum permitted 40MHz TD-LTE allocation to achieve 1Gbps in the lab last month. A combination of licenced and CBRS should be quite attractive to mainstream operators, providing a speed boost and capacity offload without becoming completely reliant on in-building CBRS systems.
Alternatively, making use of unlicenced spectrum by aggregating with 5GHz using the LAA feature should be a relatively easy way for those with less spectrum. Nokia and Megafon achieved 979Mbps using LAA in August which is close enough for me. Due to the short range of 5GHz, I’d expect this to be more popular for use indoors but this remains to be seen. There's also the option to combine standalone CBRS with 5GHz using LAA if smartphones support it.
Using higher data modulation levels
Encoding the signal with ever higher modulation rates increases the quantity of data that can be sent. This requires a better quality of radio channel, benefitting from higher performance crystal oscillators within each device as well as the fundamental design of the RF transmission.
256QAM is the highest level being used in smartphones today. Some point-to-point microwave links for backhaul use 1024QAM or even higher, but I’ve not heard of this being used in mobile devices.
Realistically, this high level of QAM won’t be supported in less than ideal radio environments, such as if penetrating a building from outside – you’ll need to have inbuilding wireless equipment to achieve that.
Deploying multiple parallel radios and antennas to process the same signals that take different routes such as being are bounced off different buildings/walls can increase signal quality and introduce diversity. This helps improve the basic RF signal quality and make it more likely a higher modulation rate can be achieved.
As I reported from MWCA last month, US operators are keen to deploy 4x4 MIMO as a minimum. This isn’t necessarily the case in other regions and certainly not for most in-building systems.
Qualcomm taking it a step further
Commercial smartphones sporting the Gigabit speed today are based on Qualcomm’s X16 LTE modem. Their latest Snapdragon X20 modem claims a further 20% increase to 1.2Gbps downlink, with uplink speeds of 360Mbps. It can handle up to 5x carrier aggregation, 4x4 MIMO and up to 12 LTE spatial streams.
One has to question how this will affect demand for 5G. We’ve seen in the past how legacy technologies have evolved so much that a step jump to radical new generations becomes less urgent. For example, the drive for IPv6 has been delayed while most of its features have been delivered through advances of IPv4.
The pace of capacity and data speed innovation with LTE suggests it has quite a long lifetime ahead of it. This makes the business case for 5G more difficult to justify.
Wi-Fi offers an alternative
It’s possible to aggregate cellular data with Wi-Fi data, using standards such as LWA and LWIP, but I’ve seen relatively little take-up of this mode. The RF performance of LTE (including LAA) exceeds that of Wi-Fi for various technical reasons, so given the choice it should provide the better technical option.
Wi-Gig, which provides Wi-Fi at 60GHz, offers multiple Gbps and the first commercial smartphone devices are becoming available this quarter. This is extremely short range (typically within the same room) but extremely high data rate. Should be ideal for quick data synchronisation or if you purchase a movie etc. Uplink speeds should be equally blazingly fast. It just won’t support mobility or be much use outdoors.
Peak rates in the lab don't always translate to live use in the field. You might get these rates when standing nearby a cellsite at 3am but don't expect headline performance all to yourself in a busy street during the daytime. Nonetheless, you should experience dramatically improved throughput of several hundred Mbps more of the time, reducing the time to load and show video clips or any larger files.
Wireless capacity is shared. The total peak data rate within any sector is only going to be available when nobody else is using it. A small share of such high rates will still be good in most cases, although might cause more variability in performance. Alternatively more small cells can be installed, giving each user a larger proportion of the available bandwidth.
Backhaul can be a bottleneck. I’m hearing that some of the busiest macrocells are being equipped with more than 1Gbps backhaul, but many still don’t. Small cells powered by CAT5 or 6 Ethernet don’t. Whether this will be noticeable by end users other than when speed testing will depend on how congested cells are.
In-building equipment required. Performance between inside and outside may become more differentiated, because these high data rates won’t be achieved using traditional outside-in techniques. Perhaps this will increase demand for properly engineered in-building cellular solutions.
End-to-end latency won’t be hugely affected. Accessing that busy or inefficient website won’t be speeded up if the end-to-end round trip delay remains high. That is being improved through internet technologies such as HTTP/2. For those wanting super low latency, this doesn’t change the fundamental 20ms frame rate of LTE.
Pricing for data services seems unaffected. Smartphones supporting Gigabit LTE are considerably pricier than the cheaper or mid-range products, so there is a premium there today. In the medium term, I would expect this feature to be incorporated into more of the mainstream devices. There’s no indication that network operators plan to charge higher rates for these faster speeds yet. Presumably they see that faster rates = higher total data consumption = larger data bundles being bought. It’s been hard to get customers to accept higher fees for higher speeds but I suspect that could be revisited in future.