5G infrastructure power supply design considerations (Part II)
In part I, we discussed the power supply design considerations applicable to the access and backhaul parts of the 5G network - the “periphery.” We learned that there were solutions for mobile terminals, small cells, masts, aggregation routers, and core routers. We also discovered that 5G brings new power supply challenges, many of which require product refinement and improvement.
In this post, we cover power supply design considerations for the core and cloud parts of the infrastructure.
Mobile cores control and distribute 5G data from user devices, acting as the “heart” or “brain” of the network. Their functions encompass aggregation of end-user traffic, session management, authentication, and security. As such, they need reliable power supplies. Failures could lead to network downtime, preventing thousands (and possibly millions) of customers from communicating with each other.
5G cores have substantial architectural changes compared to their 4G predecessors. These include:
- Cores that can accommodate the migration to millimeter-wave
- Improved network slicing capabilities for emergency services and other priority applications
- Segregation of mobility management tasks from session management functions
- Decoupling of packet gateway control and user plane functions
- Improving deliverables, such as the ability to track IoT devices
Mobile cores of 5G networks also need to support increased throughput compared to previous-generation solutions.
4G and 5G cores will coexist for many years until the 5G rollout is complete. Power supplies, therefore, will need to cater to both 5G Core (5GC) and Evolved Packet Core (EPC) functions. They will also need to accommodate the move towards “virtualized cores.” In the past, cores needed physical routers and switches to manually funnel data between network users. However, the advent of virtual cores means that much of that activity now occurs in the cloud, changing the physical power supply requirements.
Thus, in 5G, both the core and the cloud are merging. The “dumb” part of the system - the radio access network (RAN) - simply relays data traffic from users to the core. The core, on the other hand, is what allows networks to differentiate themselves from each other, and provide advanced services, such as network slicing briefly mentioned above. For the most part, the “core” is becoming more software-based. Physical 5G antennae simply feed the data they receive from users into the core - usually some local node - and then it connects with the rest of the network via cloud-connected servers.
This change is leading to some interesting implications for power supplies. Many networks want to move core functionality physically closer to users to reduce latencies. Operators, for instance, are changing server locations to reduce lag for gamers playing fast-twitch shooters on the go. They are also doing it to allow self-driving cars to communicate with each other in real-time - something that 4G infrastructure couldn’t accommodate. Power supply design, therefore, is going to change as cores become more distributed.
Based on this discussion, the rest of this post lists some of the power supply considerations that mobile core/cloud infrastructure operators need to consider.
Estimates suggest that mobile operator network operating expenses are currently growing faster than revenue, partly because of the capital costs of installing 5G infrastructure, and partly because of increased server power requirements.
Designers initially hoped that 5G would reduce operating costs by reducing overall energy consumption. However, mobile network operators now believe that they will require more local servers to provide low-latency VR, AR, IoT, and autonomy functionality for customers, upping overall energy expenditures.
"Core" cloud-native parts of the network infrastructure must have power supplies capable of supplying sufficient electricity to the following components:
- The rectifier which converts AC into usable DC
- The air conditioning apparatus to keep server microchips cool
- The backup battery system
Currently, power supply solutions deliver sufficient power to keep 4G core nodes operating. However, they may not be sufficient for 5G.
Data show, for instance, that the introduction of massive MIMO antennas will increase energy supply requirements by 1000W per sector. Plus, mobile network operators will likely have to field a greater quantity of servers to keep pace with increased traffic loads. Both factors imply that existing systems will require an upgrade.
So what are the solutions here?
Intelligent Peak Shaving
Companies supplying infrastructure in the 5G operating environment are deploying intelligent peak shaving much more widely across the grid. The idea here is to spare grid capacity and bypass AC power limits by using on-site energy storage to support peak energy usage.
Base stations, server rooms, and mobile cores require varying quantities of energy throughout the day. A typical load might be 5kW, but it could peak to over 10kW when network usage reaches its maximum.
Deriving additional power from the grid to service this demand isn’t always possible, especially if there are AC limits in place because of equipment constraints. Furthermore, utility providers may charge fees for additional power used above previously-agreed limits.
Peak shaving works by using on-site battery power to make up the shortfall. Usually, energy peaks pass quickly, meaning that there is little chance of the battery becoming depleted. Then sites can simply top up the battery when power falls back within AC limits using excess capacity.
This solution is cost-effective because many existing 4G stations already need to upgrade their battery backup systems. Additionally, it doesn’t involve any changes to the grid or fitting new and expensive rectifiers.
Power supplies will also need to become denser as the 5G network becomes more fragmented. One approach is to build units that can fit alongside all other components, such as lithium batteries, cooling systems, and rectifiers, in a single cabinet. The goal is to reduce the number of cabinets and, therefore, reduce both installation and maintenance costs.
Smart Voltage Boosting
Infrastructure architects hope that smart voltage boosting will negate the need to retrofit cables for 5G installations. Network operators are currently concerned about unacceptable voltage drops in distant base stations that could lead to a loss of service.
One solution is to retrofit old cables and increase the power to the units. However, this approach led to energy wastage and increased line installation costs.
A better option is to combine power modules with lithium-ion batteries to provide tandem power to maintain constant voltages across the network. Commentators believe that the innovation could negate the need to redo cabling and, thanks to higher voltages, may also reduce the total energy consumption of core sites.
Backup System Considerations
4G mobile core servers (and related systems) need backup electricity supplies to keep them running in the event of a power outage or other problems with the grid. In the past, mobile network operators installed bulky lead-acid cells to provide backup power as a risk-mitigation strategy. In some parts of the world, especially developing countries, these systems kicked into action frequently because of variations in the quality of the power supply. Electrical storms, bad weather, brownouts, blackouts, and voltage fluctuations occurred regularly, making battery solutions indispensable.
As discussed above, 5G load requirements are likely to rise substantially. Mobile network operators, therefore, need to rethink their backup systems and add more capacity. Old lead-acid solutions may not suffice. Newer, more reliable technologies, such as lithium-ion, should take their place.
Intelligent energy storage refers to the mixing and matching of lithium-ion cells. The idea is to combine both old and new cells and rotate them regularly to provide optimal performance. Mobile network operators are looking for ways to reduce expenditure on new batteries and, instead, develop systems that allow them to rotate batteries with different capacities and materials. On-demand battery configurations will reduce cap-ex in the short term and keep balance sheets healthier.
Load Requirement And Intelligent Power Management Considerations
Mobile network server and core load vary considerably throughout the day. Data for 4G networks shows that download speeds progressively slow in the afternoon and reach a minimum by around 9 pm, before speeding up again as nighttime falls. Therefore, industrial computer power supplies need to be able to react intelligently to changing load requirements in the 5G core.
Importantly, power supplies must be able to accommodate surges in demand from processors and other components in the servers without damaging any of their components in the process. Ideally, power supplies should supply at 150 percent of their rated power to accommodate spikes in 5G network demand. Such in-built capacity could help to prevent momentary network stoppages or unacceptably high IoT latencies.
Mobile network operators (MNO) (and other companies tasked with maintaining physical hardware) also need to consider power supply mounting - how exactly power supplies are going to fit inside server cases and cabinets.
While standard ATX mounting is common in regular computers, it isn’t always appropriate for industrial applications, such as mobile networks. Fortunately, there are several other configurations available to meet MNO needs.
Small form factor power supplies, for instance, are ideal for MNOs looking for power supply solutions in space-constrained environments, such as server and base station cabinets. Smaller power supplies naturally fit into these systems.
Thin form factor 12V power supplies fit FlexATX and MicroATX layouts. Their long and narrow shape means that they are suitable for low profile systems while also retaining vital internal components. Cooling fans let them exhaust tool air out of the back away from hot-running equipment, ideal for 5G’s fragmented and virtualized core infrastructure.
Thermal Management Considerations
While many 5G core and server facilities use air-conditioned rooms to keep components cool, they still require thermally-efficient solutions to keep costs down.
Ambient temperatures play a role in the life of power supply units. Data suggest that small reductions in the ambient temperatures of power supplies of 10° C or less can double capacitor lifetime. Power supplies begin to derate (operate below their stated specifications) as temperatures rise. AC/DC power supplies, for instance, begin to derate around 50° C and lose around half their rated load at 70° C. Hence, keeping temperatures low allows engineers to access additional capacity.
The position and orientation of power supply units are important too. In some situations, 5G network infrastructure engineers will install open-frame power supply units. These solutions do away with the shroud and accompanying fan and simply sit alongside other components, dissipating heat passively.
Without a shroud, their effect on the surrounding components depends on:
- The available system cooling (whether the server has onboard fans)
- The orientation of the unit
- The unit’s mounting position inside the case
- The load being applied
While power supply units aren’t the hottest-running components in a typical server, excess heat could affect the performance of the overall unit, causing it to fall outside of its intended operating parameters. That, in turn, could adversely affect network performance, leading to latencies and packet delays. If open frame enclosures produce too much heat then designers may have to replace them with enclosed systems with fans.
Electromagnetic Interference Considerations
Mobile network power supplies operating in high electromagnetic interference (EMI) environments often need substantial protection from ambient noise. Wholesale and retail open frame power supplies do not have enclosures to protect them and so are more suited to remote server farms in the cloud, not base stations or units in the radio access network.
Power supplies come with a range of technologies to protect them against electromagnetic interference. These include:
- Distancing input and output wiring
- Mounting open frame power supplies on metal sheets
- Using U-channel enclosures
- Using an adapter
Deploying these features can help in 5G applications where EMI is a risk.
5G rollout presents new and interesting challenges for power supply design. Engineers must consider efficiency, load, noise thermal management, and how to integrate power supplies with backup systems. They must also install power supplies capable of adapting to the idiosyncrasies of the 5G RAN and core, including smaller form factors and narrower voltage tolerances.
FSP solves many of the power challenges for firms rolling out 5G and provides a comprehensive network of support.
- Modular product design. We make it easy to create power supply solutions across both the core and RAN 5G network, no matter the location of devices.
- Digitized product design. FSP’s digitized product design makes it easy to provide power supplies that meet modified standards.
- Quick response to customer demands. We provide power supplies that cater to variable operating environments, plus strategic-level solutions for the entire stack.
- Intelligent power supplies. Our units are high-power-density, high-efficiency, and able to withstand the harsh environmental requirements of 5G.
- High quality units. FSP provides high-quality products that meet the needs and expectations of engineers involved in the 5G rollout.