Create the hype, but don’t ever believe it. – Simon Cowell
Despite the tons of early hype, the worldwide 5G race has just begun. In a race to announce the first operational 5G network in the US, cellular communications vendors AT&T and Verizon have both initiated limited 5G cellular trials in a handful of cities. AT&T’s 5G trials started in twelve cities along the East coast of the US and will spread to San Francisco and San Jose later in 2019. Verizon’s trial cities include Sacramento, Los Angeles, Houston, and Indianapolis. Many of these 5G announcements increasingly discuss 5G cellular communications for IoT connectivity, because potentially selling cellular subscriptions for perhaps a trillion IoT devices is orders of magnitude more interesting than selling them to the seven billion people living on the planet, even if those IoT devices pay a lower monthly fee. You make it up in volume.
Here are some early “5G IoT” examples:
- Nokia has posted a “5G demonstration” on YouTube that shows three robots controlling a tilting tray to stop a rolling ball dead in its tracks. It’s quite a real time balancing act. The demo supposedly compares 5G and 4G performance for an IoT application, but it’s really just a demonstration of low-latency Ethernet. No 5G was involved in this demo, except in the theoretical sense.
- Korea Telecom (KT) opened a café near Seoul Subway’s Gangnam Station that features a robotic barista. There are no human employees. A local dal.komm COFFEE franchise operates the café. KT’s robotic barista takes orders, makes and serves cappuccinos, and maintains an IoT connection that carries video, control, and status communications to a “centralized dal.komm COFFEE control room” over a 5G wireless connection. (Coffee needs a control room?)
- More recently, a doctor in China’s Fujian province demonstrated liver surgery on a lab animal (a pig) using a remotely controlled a surgical robot connected via a low latency 5G wireless connection.
These announcements and demonstrations imply that 5G cellular connections for IoT services are imminent and that they will soon replace 4G LTE cellular communications. Neither of those assertions is especially true. Here’s what is true: 5G communications standards for mobile phone communications are not yet finalized, and the 5G standards for IoT applications have not yet been announced. Meanwhile, 4G LTE communication technology – including cellular services, hardware, and software – already delivers reliable IoT connectivity over long distances and will continue to do so for many years into the future.
On January 8, Jonathan Wells, Managing Partner of AJIS Consulting and wireless consultant, gave a comprehensive “State of 5G” presentation at the monthly IEEE-CNSV (Consultants Network of Silicon Valley) meeting. During his presentation, Wells touched on the 5G topics listed above and many more. What follows are some of the specifics Wells discussed in his talk:
5G cellular communications seeks to improve upon 4G specifications in several dimensions:
- Latency: 4G LTE communications latencies are on the order of 30 to 50 milliseconds. 5G advocates hope to achieve latencies on the order of one millisecond. That’s a 30x to 50x latency improvement. Anyone wishing to perform remote robotic surgery will need this improved latency. Most IoT applications can tolerate current 4G LTE latencies.
- Bandwidth: At its best, with optimal conditions, a 4G LTE connection can deliver 100 megabits per second of bandwidth, which is more than sufficient for most IoT applications. The goal for a 5G connection is 10 gigabits per second. That’s a 10x bandwidth improvement, but it’s not clear how one 5G connection might use all of that bandwidth.
- Mobility: Currently, a 4G LTE network can accommodate a mobile connection traveling at 350 kilometers per hour. Again, this is more than sufficient for most IoT applications except those associated with the fastest high-speed trains or jet aircraft. The 5G goal is 500 kilometers per hour, which is a modest increase and quite achievable, but it’s still not fast enough to keep up with jet travel.
- Connections: 4G LTE networks can accommodate approximately 10,000 connections per square kilometer. That’s sufficient for most mobile handset densities and for current IoT applications, but when the IoT becomes ubiquitous, when those trillion IoT devices suddenly appear, more connection density will be required. The 5G connectivity goal is one million connections per square kilometer to network all kinds of “things” from vehicles, to refrigerators, to thermostats. That’s a 100x increase in the network connection density. To achieve this, 5G networks will likely require many more base stations per square kilometer. (No Virginia, there are no 500kph refrigerators. Thankfully.)
5G Standards for IoT Communications
All cellular networks rely on standards to ensure interoperability of base stations and terminal equipment. Several standards bodies are involved in the creation of these network standards including:
- The ITU (International Telecommunication Union): The ITU was founded as the “International Telegraph Union” in 1865. It’s the oldest specialized agency of the United Nations and it’s the overarching standards body that manages shared use of the world’s radio spectrum, including cellular radio frequencies. The ITU also defines the cellular phone generations. More specifically, the ITU’s Working Party 5D defined what a 5G network will be in 2020 in a yet-to-be-issued specification called IMT-2020. The ITU also certifies cellular networks for conformance against the specs. All 193 UN member countries are members of the ITU.
The ITU is currently taking 5G recommendations from other worldwide standards bodies and building a consensus decision for the IMT-2020, which will be issued in 2020. The ITU holds World Radio Conferences (WRCs) every three to four years to build final consensus, and the next WRC (WRC-19) will be held in October and November, 2019 in Sharm el-Sheikh, Egypt. No cellular network can be officially called 5G until there’s an IMT-2020 spec. However, that has not stopped cellular operators, including AT&T and Verizon, from pre-announcing their 5G networks.
- The Third Generation Partnership Project (3GPP): The 3GPP originally set the definitions for 3G mobile phones. First- and second-generation cellular phones varied from country to country and there was no single, worldwide standard for these phone generations. First-generation cellular phones were analog phones. Second-generation cellular phones went digital, but there were two digital standards used in the world: IS-95 based on CDMA for the US and GSM for the rest of the world.
The 3GPP was formed in 1999 to develop 3G cellular specifications that would work worldwide. Since then, the 3GPP has released a new set of specs about once a year, as its standards evolve. The 3GPP Release 99 specification defines 3G phones today. After developing Release 99 for 3G, the 3GPP developed standards for 4G LTE networks. It is now working on standards for 5G cellular phones. (It seems that the 3GPP likes to evolve standards but not its own name.)
Interest in 5G standards was so high that the 3GPP partitioned its 5G specifications into what it calls 5G Release 15 and 16. The current version is Release 15, which is being released in three “drops” – the Early drop, the Main drop, and the Late drop – appearing at roughly semi-annual intervals.
The Early drop of Release 15 appeared in March, 2018. This drop allowed vendors to start developing prototype 5G equipment. The 3GPP’s Release 15 Main drop appeared in September, 2018. It did not differ significantly from the Early drop, so vendors could simply make software updates to bring their prototype 5G equipment into compliance with the Main drop. The Late drop won’t appear until June, 2019. That is when 5G specs will freeze and vendors can design equipment and services against real standards. Relevant to the IoT discussion, the first two Release 15 drops focused on data rates and spectral efficiency. The Late drop focuses on improving latency. If you want to operate on pigs, balance balls in real time, or serve coffee without spilling it on the customers, you’ll probably need the Late drop’s improved latency.
- Verizon’s 5G TF: Verizon was impatient to move into 5G, so the company established its own standards group with ecosystem partners including Cisco, Ericsson, Intel, LG, Nokia, Qualcomm, and Samsung. That group is called the 5G Technology Forum (TF). Verizon is using these 5G TF specifications to facilitate its early 5G trials.
Why not 5G?
Despite all of this effort, 5G network and equipment standards are not yet finalized. Even when the standards are finalized, it will take several years for standardized 5G networks and equipment to be built. Meanwhile, 4G LTE networks have been in commercial operation since 2011 and are quite capable of delivering nearly all of the performance required by most IoT applications.
By at least one estimate, 4G LTE networks have about 40% of the current, worldwide cellular market, while the older 2G and 3G networks represent about 30% of the market each. Even by 2025, 5G market penetration is not expected to be more than about 15%. Clearly, today’s IoT designs that need the long distance capabilities of a cellular network should conform to the 4G LTE standards, or even earlier cellular standards, which are already in place and are widely used.
The 3GPP’s Release 13 of the LTE standard defined new narrowband LTE categories for IoT applications: category M1 (Cat-M1), formerly known as eMTC, and category NB1 (Cat-NB1), formerly known as NB-IoT. These new categories extend LTE by enabling support for lower data rate IoT applications. Cat M1 defines a 1.4 MHz channel size with a throughput of about 375 kilobits per second for uplink and 300 kilobits per second for a downlink. Cat NB1 defines a much smaller channel size of 200 kHz with a throughput measured in tens of kilobits per second. Cat M1 latency is approximately 10 to 15 milliseconds. Cat NB1 latencies are measured in seconds and can be as much as much as 10 seconds in some deployment scenarios.
Although these data rates and latencies are not sufficient for carrying video, as employed by KT’s robotic barista, for surgical telepresence, or for communications among autonomous vehicles, these rates are ample for many sensing IoT applications such as meter readers, health status monitors, and highly mobile fitness applications that can benefit from the long reach and ubiquitous presence of cellular communications. Currently, and for the foreseeable future, no other low power, wide area, wireless technology offers the scalability, security, and longevity of the established 4G LTE networks.
Five Takeaways for 5G
Throughout his 5G talk, Wells dropped takeaway points that he wished to emphasize. He provided five points in all:
- Most of what you hear about 5G is hype.
- The value proposition for 5G is unclear.
- The term “5G” has no technical meaning (sort of). There is no 5G standard at the moment.
- 5G cellular communications will be ultrafast, but the rollout of 5G networks will be anything but.
- Most of the technologies hyped as “made possible only by 5G” applications don’t actually need 5G networks.
So, from Wells’s perspective (and he certainly convinced me), 5G is still a few years off. However, today we have 4G LTE.