Working from home may have reduced the stress of commuting, but it put a heavy strain on the power grid and ate into energy reserves. Still, with utilities such as electricity and water, it’s possible to buy or borrow from adjacent grids or territories.
The same cannot be said for internet service providers. They cannot just borrow from another service provider. Instead, their customers would have to rely on coffee house internet or tether their cellular devices. As hybrid work becomes a mainstay, ISPs need to create contingencies during outages.
So, is there a way to make the internet more like a public utility, broadly available, affordable, and reliable with failover capability? One emerging technology that has this potential is 5G-based satellite internet. The industry is working on standardized internet protocols that could enable 5G-internet providers to pool resources and “borrow” bandwidth from each other when needed. As companies build and deploy 5G satellites, they collectively create a comprehensive array that blankets many parts of the world with 5G internet. If company A’s satellite malfunctions, it could rely on company B’s adjacent satellite to prevent outages, for a premium, of course. This sharing model is analogous to how cell phone providers deliver cellular service while users travel. Devices connect to local towers, and users receive a roaming charge.
Space-based silicon and even 5G-based satellite internet are still in their infancy. Companies face two significant hurdles before they can begin building and deploying satellites at volume: frozen 5G-based internet standards and radiation-hardened silicon design techniques. The 3GPP (3rd Generation Partnership Project) release 17 (Rel-17) unlocks half of the problem for standardization past the stratosphere. The new 5G standard builds on the early work set forth by Rel-15, where 3GPP defined the minimum set of parameters for non-terrestrial networks. Rel-16 took these parameters and built the framework for higher- and physical-layer protocols.
With the latest hardened iteration of 5G, the standards board solves timing issues and improves the quality of service for NTNs. Satellites use GPS to calculate round-trip time (RTT) to ensure that it receives the signals at the right frequency. Rel-17 refines these calculations to apply a buffer for the Doppler frequency. These devices also use the timing information to generate an offset between downlink and uplink transmissions.
To improve the quality of service for low Earth orbit (LEO) satellites, 3GPP expands the number of stop-and-wait HARQ processes to 32. For higher satellites, 16 HARQ processes are not enough and created stalls while waiting for feedback. Additionally, the fewer processes create switching issues for satellites or endpoints in motion. As devices move, HARQ processes stall. After a predetermined number of stalls, the endpoint device will search for a new satellite, a power-hungry and time-consuming activity. Doubling the number of processes reduces the frequency of these stalls and improves data throughput for LEO satellites.
While Rel-17 catalyzes new architectures for space-based silicon, advancements in rad-hard technology will create more reliable space-bound silicon. In space, silicon gets bombarded with ionizing radiation, which causes single event effects (SEEs). SEEs are traditionally transient and occur when ionizing radiation hits silicon and bores a nanoscopic tunnel. This tunnel may cause bit flips, sudden timing changes, or voltage shifts. Transistors may permanently stay open or closed if a device’s insulating layer accumulates enough ionized particles, causing permanent device failures.
We experience ionizing radiation everywhere in relatively small and harmless doses (0.0062 Sv/year). LEO devices, however, experience anywhere from 1 to 100 Sv per year, depending on the latitude. The massive difference in radiation comes from solar flares, cosmic rays interacting with our atmosphere, and ionized particles from deep space.
Analog-based PLLs are especially susceptible to radiation damage. Large accumulated doses of ionization cause a shift in threshold voltages, altering the PLL’s charge pump bias currents. These parameter shifts decrease closed-loop jitter performance and loop acquisition behavior, creating an unreliable clock signal that can propagate race conditions throughout downstream logic. Conventional PLLs are also vulnerable to SEEs. If ionizing radiation strikes the frequency divider or phase detector, the PLL will lose lock and will have to relock. The chip will be inoperable while the PLL relocks, causing transient system failures.
Movellus’ clock generation IP solutions use standard cells with rad-hard counterparts. Unlike their analog counterparts, digital clock generators are far more robust. Hardening techniques are integrated at the circuit and architectural level through a digital approach.
There are several things that can be done to make chips more robust and reliable:
- At the circuit level, rad-hard standard cells are used for each component in a clock network, including clock gates, inverters, and buffers.
- For storage cells, fail-safes are added, which revert to previous known states in the event of an ionizing strike.
- Clock generation redundancy can be expanded to the architectural level, adding combinatorial logic into the clock network IP while adjudicating signals through a voting multiplexer. In Movellus’ case, these techniques allow IP to withstand ionizing doses greater than 1kSv and reduce clock-related SEEs to 10-12 errors/bit-day.
In the event of a clock generation failure, our IP locks in 100 reference clock cycles, mitigating downtime for critical workloads. Additionally, the IP integrates SEE filters on vital diagnostic signals that quickly switch modes (SCAN, functional, reset) to help users isolate failures.
To have their silicon reach the stars and beyond, companies need reliable IP, including clock network IP. Through our recently announced strategic partnership with BAE, Movellus will continue to tailor its clock generation IP for the next generation of space-bound silicon.
It is only natural that our architectural renaissance will expand past earth into space. Standardization and radiation-hardened design techniques are quickly catching up to enable star-bound silicon and deploy 5G-based satellite internet. It has the potential to make the internet more like a utility, improving the quality of service for many customers and expanding coverage zones past the suburbs. Given its far reach, rural communities, offshore and arctic research, and disaster zones will also benefit from enhanced internet access beamed down from the stars.
This article originally appeared on Semiconductor Engineering