5G (or 5G-NR for new radio) is the next generation IP-based communication standard being drafted and designed to replace 4G-LTE. That said, it draws on some technologies of 4G-LTE but comes with substantial differences and new capabilities. So much material has been written on 5G that it dwarfs even current Cat-1 and Cat-M1 material. In part, 5G promises to deliver substantial abilities for IoT, commercial, mobile, and vehicular use cases. 5G also improves bandwidth, latency, density, and user cost.  Rather than build different cellular services and categories for each use case, 5G attempts to be a single umbrella standard to serve them all. Meanwhile, 4G-LTE will continue to be the predominant technology for cellular coverage and will continue to evolve. 5G is not a continuing evolution of 4G; it derives from 4G but is a new set of technologies. In this section, we will only discuss elements that pertain to IoT use cases or are of merit and have the potential to become part of the 5G specification.

5G is aiming for initial customer launch in 2020, however, mass deployment and adoption may follow years later in the mid-2020s. The goals and architecture of 5G are still evolving, and have been since 2012. There have been three distinct and different goals for 5G: http://www.gsmhistory.com/5g/. Goals range from converged fiber and cellular infrastructure, ultra-fast mobiles using small cells, and lowered cost barriers of mobile, all part of the 5G story. Again, the ITU-R has approved the international specifications and standards, while the 3GPP is following with a set of standards to match the ITU-R timeline. The 3GPP RAN has already begun analyzing study items as of release 14. The intent is to produce a two-phase release of 5G technologies. Phase one will complete in 2018 with release 15 and phase two is targeted for release 19 in 2019. Both align with the 2020 goal of commercial launch.

The general consensus of features in 5G that should be understood and have a good prospect of being in the final release are:

1. Enhanced Mobile Broadband (eMBB):
   • 1 to 10 GBps connections to UEs/endpoints in the field (not theoretical)
   • 100% coverage across the globe (or perception of)
   • 10 to 100x the number of connected devices over 4G-LTE
   • Connectivity at a speed of 500 km/h
2. Ultra-Reliable and Low-Latency Communications (URLLC):
   • Sub < 1 ms end-to-end round-trip latency
   • 99.999% availability (or perception of)
3. Massive Machine Type Communications (mMTC):
   • 1000x bandwidth per unit area; this implies roughly 1 million nodes in 1 km²
   • Up to a 10 year battery life on endpoint IoT nodes
   • 90% reduction in network energy usage:

Current 4G-LTE systems mainly use frequencies below the 3 GHz range. 5G will radically change the spectrum usage. While the space below 3 GHz is significantly congested and sliced into slivers of bandwidth, 5G may use a multitude of frequencies. Under strong consideration is the use of millimeter waves (mmWave) in the unlicensed 24 to 100 GHz range. These frequencies directly address Shannon's Law by increasing the bandwidth B of the law with extremely wide channels. Since the mmWave space is not saturated or sliced up by various regulatory bodies, channels as wide as 100 MHz in the 30 GHz to 60 GHz frequencies are possible. This will provide the technology to support multi-gigabit per second speeds.

The principal issues with mmWave technology are free space path loss, attenuation, and penetration. If we remember that free space path loss can be calculated as L_fs = 32.4 + 20log₁₀f + 20log₁₀R (where f is the frequency and R is the range), then we can see how the loss is affected by a 2.4, 30, and 60 GHz signal as:

• 2.4 GHz, 100 m range: 80.1 dB
• 30 GHz, 100 m range: 102.0 dB
• 60 GHz, 100 m range: 108.0 dB 
• 2.4 GHz, 1 km range: 100.1 dB
• 30 GHz, 1 km range: 122.0 dB
• 60 GHz, 1 km range: 128.0 dB

20 dB is significant, but with mmWave, antennas can accommodate significantly more antenna elements than a 2.4 GHz antenna. Free path loss is significant only if the antenna gain is independent of frequency. If we keep the antenna area constant, it is possible to mitigate the effects of path loss. This requires Massive-MIMO (M-MIMO) technology. M-MIMO will incorporate macrocell towers with 256 to 1024 antennas. Beamforming at the macrocell will be used as well. When combined with mm-Waves, M-MIMO has challenges in terms of contamination from nearby towers, and multiplexing protocols like TDD will need to be re-architected.

Attenuation is a very significant issue. At 60 GHz, a signal will be absorbed by oxygen in the atmosphere. Even vegetation and the human body itself will have a serious effect on signals. A human body will absorb so much RF energy at 60 Ghz that it will form a shadow. A signal traveling at 60 GHz will have a loss of 15 dB/km. Thus, long-range communication will be subpar using 5G at 60 GHz and will require either a blanket of small cells or a drop to a slower frequency space. This is one of the reasons the 5G architecture will require multiple bands, small cells, macrocells, and a heterogeneous network. 

Finally, material penetration in the mmWave spectrum is challenging. mmWave signals attenuate through drywall at 15 dB and glass windows on buildings contribute to a 40 dB loss. Therefore, indoor coverage with a macrocell is close to impossible. This and other types of signaling issues will be mitigated through the widespread use of indoor small cells:

UEs may use multiple bands simultaneously. For example, an endpoint device may use lower frequencies for long-range communication and switch to mmWave for indoor and personal communication. Another scheme being considered is Dual Connectivity. Dual Connectivity steers data traffic across multiple bands dependent on data type. For example, the control plane and user plane of the stack are already divided. User plane data may steer to a nearby small cell using a 30 GHz frequency while control data may be steered to a long-range eNodeB macrocell tower at the typical 4 Ghz rate.  

Another improvement for increased speed is spectral efficiency. The 3GPP is focusing on certain design rules:

• 15 kHz subcarrier spacing to improve multiplexing efficiency.
• Flexible and scalable numerology of symbols from 2^M symbols to 1 symbol to reduce latency.

As previously discussed, spectral efficiency is given as bps/Hz, using D2D and M-MIMO to improve spectral efficiency along with changes to the air interface and new radio. 4G-LTE uses OFDM, which works well for large data transmissions. However, for IoT and mMTC, the packets are much smaller. The overhead for OFDM also impacts latency in very dense IoT deployments. Hence, new waveforms are being architected for consideration:

• Non-orthogonal Multiple Access (NOMA): Allows multiple users to share a wireless medium.
• Filter Bank Multi-Carrier (FBMC): Controls the shape of subcarrier signals to remove side-lobes through DSPs.
• Sparse Coded Multiple Access (SCMA): Allows data to be mapped to different code from different codebooks.

Latency reduction is also a goal of the ITU and 3GPP. Latency reduction is crucial for 5G use cases such as interactive entertainment and virtual reality headsets, but also for industrial automation. However, it plays a large role in power reduction (another ITU goal). 4G-LTE can have a latency of up to 15 ms on 1 ms subframes. 5G is preparing for sub-1 ms latency. This too will be accomplished using small cells to route rather than congested macrocells. The architecture also plans for Device to Device (D2D) communication, essentially taking the cell infrastructure out of the data path for communication between UEs.

4G systems will continue to exist, as the rollout of 5G will take several years. There will be a period of coexistence that needs to be established. Release 15 will add further definitions to the overall architecture such as channel and frequency choices. From an IoT architect perspective, 5G is a technology to watch and plan for. IoT devices may predicate a WAN that will need to operate for a dozen years or more in the field. For a good perspective on the key points, constraints, and the detailed design of 5G, readers should refer to 5G: A Tutorial Overview of Standards, Trials, Challenges, Deployment, and Practice, by M. Shafi et al., in IEEE Journal on Selected Areas in Communications, vol. 35, no. 6, pp. 1201-1221, June 2017.