When we talk about 5G wireless, we tend to think about smartphones, and to focus on the high-speed broadband that 5G can provide. But smartphones are just one of the many use cases for 5G.
As well as increased performance, 5G offers other features such as low latency, low power consumption, good security and network slicing. This means that it is an attractive technology for a wide range of applications, including the IoT, wearables, VR/AR headsets and the metaverse, surveillance, healthcare, industrial, automotive and many more.
In fact, industrial machine-to-machine (M2M) applications are forecast to account for 70% of non-handset volumes in 5G by 2026, especially in China, while the Ericsson Mobility Report predicts that there will be 5.5 billion cellular IoT connections in 2027, up from 1.9 billion in 2021.
Figure 1: 5G applications
For these other applications, most of them require embedded wireless products, that can provide a balance between sufficient performance, low latency (URLLC – see Figure 1) and low power consumption. They often don’t need the high speeds of 5G’s eMBB (enhanced mobile broadband) capabilities, perhaps just sending small packets of data from a sensor or edge device. Any particular application will have its own set of requirements, which means flexibility is needed to provide an optimized solution in each case.
In the past, cellular broadband would not have been able to provide the right solution for many of these use cases, as the products available used too much power and were not compact enough. Instead, the preferred technologies were often Wi-Fi or Bluetooth – which both still have their place as alternatives alongside cellular, but cannot provide all the features and performance of 5G, or the ability to connect to public networks.
This is set to change with the development of 5G RedCap (Reduced Capability). Also known as NR-Lite, the new 5G RedCap standard is going to be hugely important for non-handset applications. It provides performance comparable to LTE Cat. 4, around 85 Mbps (for a single antenna/layer mode), while offering improved latency, and other 5G features such as positioning (huge consumer potential in personal trackers), mmWave and unlicenses spectrum, network slicing, etc. RedCap enables simpler solutions to be used than ‘standard’ 5G, with reduced power consumption, thus enabling low cost 5G deployments. RedCap promises a new option of baseband integration into SoCs, with cost effective RFIC integration, supporting reduced complexity true half duplex operating mode.
Table 1: RedCap compared to legacy cellular technologies
RedCap is expected to be used widely in use cases such as industrial sensors, surveillance cameras (both for smart cities and home security) and in wearables. Its specification will be finalized in 5G Release 17, expected to be completed in 2022. The first RedCap modules are likely to be available in late 2023 or 2024, and RedCap will start picking up substantial market share in the second half of this decade.
Once they have decided that 5G is the right technology for them, semiconductor companies and OEMs need to source or develop the appropriate components. Specifically, the 5G modem is an important part of the overall system, which greatly affects the performance, cost and power consumption of the 5G device being designed.
5G modems are available as off-the-shelf parts from incumbent vendors such as Qualcomm. However, these modem chips can be relatively expensive, and do not have a good integration path into SoCs for cost reduction.
Instead, an OEM may want to develop application-specific 5G modems in-house. As well as keeping costs low, which is necessary for many price-sensitive applications, this will enable them to optimize the 5G modem for their own specific requirements. By creating their own 5G modem, companies can choose exactly where they want to compromise between performance, power consumption and latency. OEMs can also integrate 5G baseband into a connectivity SoC, in the example of Figure 2.
Figure 2: 5G RedCap as part of set of connectivity options for wearables
With form factors getting smaller and smaller, particularly for the IoT, achieving a compact solution is also important. An attractive answer is to use a System on Chip (SoC), with intellectual property (IP) from one or more suppliers integrated into a single chip solution – compared to the multi-chip solution that would be needed with a standard off-the-shelf 5G chips.
That’s all well and good, but developing 5G modems is a difficult, complicated task. In the past, companies could start with a DSP core, and then implement the modem primarily in software. Nowadays, that’s not good enough – the power consumption of this approach is too high to be usable for most use cases. To deliver high enough efficiency, you need to undertake many optimizations, and accelerate many of the baseband operations in dedicated HW accelerators, which can be an overwhelming task for most product design teams.
Flexible platform approach
So far, we’ve looked at two options for the 5G modem. Firstly, buying a dedicated modem chip, which may be costly and inflexible as well as meaning you’re dependent on the roadmap of an external supplier – thus making long-term planning difficult, and increasing the risk of changes that negatively affect your design. Alternatively, you can devote lots of resources to in-house development, but this will have a big impact on time to market, and increases risk.
There is a third way: using a flexible hardware and software platform which can be integrated into your own SoC. One of the key benefits of this kind of approach – such as CEVA’s PentaG2, a full IP platform that combines DSPs with accelerators for baseband processing – is flexibility. As well as enabling a high-performance yet power-efficient modem to be designed in a cost-effective way, this is important because some of the technologies involved, such as RedCap, are not yet fully standardized. This means any solution may potentially need to be re-worked to meet future changes in the 3GPP specifications, thus HW/SW partitioning is critical, as well as using configurable and flexible HW accelerators.
With a flexible, IP-based platform, it’s also possible for companies to bring in their own proprietary algorithms and IP to run on the SoC, such as channel estimation, forward error correction (FEC) or advanced equalization. This can be in addition to the main IP blocks provided with the platform, or to replace one or more of the platform’s standard accelerators.
The amount of modification needed does depend on the application. For example, AR/VR headsets need to provide very low latency, and excellent quality of service (QoS), to maintain a good user experience – and 5G is the only technology that can meet these requirements. As a portable consumer device, there are also likely to be tight constraints on cost and power consumption for most AR/VR products. To meet all these conflicting demands, the design team is likely to have to extensively customize the 5G modem in their product.
To develop 5G modems, the two main options so far have been to buy an off-the-shelf chip from a modem supplier, or handle development in-house. Both have drawbacks: 5G modem chips can be costly, and you cannot optimize their design. Alternatively, doing everything yourself can be slow, expensive and risky, assuming you can even hire enough of the right people.
Instead, using a flexible hardware/software platform such as PentaG2, substantially reduces the entry barriers for semiconductor companies and OEMs looking to handle 5G modem design for themselves, whether for the new markets opening up such as the IoT or simply for 5G handsets. With this platform approach, designers can integrate their 5G modem into an SoC, and take advantage of its scalability and flexibility to optimize their 5G design for their specific requirements.
Published by EE times.