According to Mike Hibbett, Business Development Manager with Firmwave, a Dublin-based technology design company, “today’s technology allows us to reconsider applications that were previously impractical”. What Hibbett modestly underplays is that it still takes clever engineers to pick which advanced technology can be best applied to solve a previously intractable engineering problem. And to choose Bluetooth Low Energy (Bluetooth LE), an RF technology designed for low-cost, short range wireless Internet of Things (IoT) applications, as the basis of a low-Earth orbit (LEO) satellite broadcast system borders on genius.
Firmwave is based at DCU Alpha, an innovation campus at Dublin City University (DCU), and specializes in the design of cellular and Bluetooth LE IoT solutions such as sensors and gateways. DCU Alpha is also home to the European Space Agency (ESA) Maker-Space for Satellite Communication, the agency’s offshoot that encourages commercial companies to help out with space research.
Back in early 2018, an invitation to tender from ESA Maker-Space fell on Hibbett’s desk. The challenge issued by the organization was to enable it to “maximize its spaceborne assets,” in other words, increase utilization of satellites. Specifically, ESA Maker-Space asked companies to put together a proposal to use a satellite to transmit Bluetooth LE signals down to Earth and then rebroadcast them using beacons.
A tough challenge
Among other tasks, satellites operate as radio relays. Earth- bound transmitters are limited by obstacles such as mountains and the planet’s curvature, but by beaming the message up to a satellite and then using the onboard radio to retransmit, vast swathes of the planet can be serviced from one spacecraft.
ESA Maker-Space realized that if Bluetooth LE signals could be sent via satellite, inexpensive battery-powered receivers could be widely distributed and used to send messages via Bluetooth beacons to smartphones in close proximity. Such a system would form a cost-effective method of conveying information. ESA just needed the help of some innovative engineers to put the idea into practice.
It’s not a trivial engineering problem. Bluetooth LE radios operate in the 2.4-GHz industrial, Scientific, and Medical (ISM) band which sits between 2.4 and 2.483 GHz. The signals from some satellites are broadcast in the ‘S band’ between 2.488 GHz and 2.495 GHz. Further, while some satellites carry radios with a transmit power exceeding 90 dBm, once that signal travels down to Earth, signal strength can be reduced to -77 dBm.
Another complication is the effect of an LEO satellite’s velocity. Orbital speed is determined by how far the satellite is from the ground; the lower the craft flies, the faster it has to travel to stay aloft. This velocity imparts Doppler shift on the radio signal affecting its frequency.
ESA’s specification called for a system that could communicate with a satellite orbiting at a height of between 160 and 2500 km. At 160 km, the orbital velocity is around 7.8 km/s and Doppler shift adds up to 60 kHz to the signal frequency as the satellite nears and decreases it by up to a 60 kHz as it moves away. The Doppler effect has to be compensated by the radio software, otherwise the signal will be lost.
Finally, when making “non-connected” Bluetooth LE transmissions (i.e. one-way, using the protocol’s three advertising channels), the Bluetooth LE RF software (the "stack") blocks out-of-band signals and continually hops between the three frequencies. Changing this behavior demands coding changes deep in the stack.
Firmwave’s engineers demonstrated their expertise by coming up with practical solutions to these tough engineering challenges. Initially the team narrowed down the Bluetooth LE chip selection to Nordic’s nRF52832 System-on- Chip (SoC) which, despite being primarily designed for 2.4 GHz ISM band operation, is capable of receiving an RF signal of up to 2.5 GHz. Moreover, the chip features a receive sensitivity of -96 dBm – providing a reasonable margin to receive a weakened signal even when the satellite is orbiting at the high limit of the ESA specification while being low on the horizon.
However, the Nordic SoC’s nominal receive sensitivity assumes the use of a linear polarized antenna. These are optimized for picking up signals typical of Bluetooth LE applications (i.e. transmitted in a horizontal plane over short ranges) but are not so good for the right hand circular polarized (RHCP) S band transmissions coming from overhead. Firmwave’s team optimized sensitivity by matching the SoC with an antenna designed specifically for RHCP signals.
There was also a need to modify some of the Bluetooth LE stack code to make the satellite link work. The key change was to alter the operating frequency to 2.488 GHz to suit the S band transmissions and then to prevent the protocol performing Bluetooth tech’s conventional frequency hopping so that it remained locked to the single channel. Other protocol changes included disabling Bluetooth LE tech’s cyclic redundancy check (CRC) and modifying advertising packets. The engineers then wrote some firmware to compensate for Doppler shift however fast the satellite was moving and whatever its orbital position.
The Firmwave team quickly realized that engineering a single Bluetooth LE SoC module to perform both the satellite reception and beacon transmission would be impractical. Such an arrangement would require running two Bluetooth stacks, one modified for the former task and a conventional stack to deal with the latter. Although the Nordic chip is capable of concurrently running two stacks, Firmwave took the far simpler route of employing two separate modules (based on Nordic nRF52 Series SoCs) each dedicated to a specific task. Because the modules sit adjacent they can be linked via a wired connection between each SoC’s UART port.
“Adding a second module does increase solution cost,” explains Hibbett. “But with high-quality, off-the-shelf modules like the Raytac [nRF52832] product we used costing just $3.50, it is a small price to pay for shortened development time and increased flexibility of the end product. Some additional cost could be shaved in commercial applications by employing a module with an nRF52810. This is a good option because the chip is ideally suited to beacon applications but less expensive than an nRF52832.
“The modules are certified products and because the one using the modified Bluetooth LE stack only receives, it requires no re-certification. The second module is unmodified so, of course, no additional certification is needed,” explains Hibbett.
To prove the concept, the Firmwave team used an Earthbound software defined radio (SDR) — which could accurately emulate the strength and Doppler shift of the LEO satellite’s transmissions— and two Nordic nRF52 Development Kits (DKs). One DK acted as the satellite receiver and the other as a Bluetooth (Eddystone) beacon.
Hibbett has worked with Nordic’s wireless technology for over a decade and respects the company’s ultra low power wireless technology heritage. But while it’s a leader, Nordic isn’t the only supplier of Bluetooth LE SoCs; so why did Firmwave choose the company’s products for this application?
“There were four key reasons,” explains Hibbett. “First and perhaps most significantly, Nordic’s nRF52832’s datasheet notes that the radio can operate up to 2.5 GHz, sufficient to pick up the part of the S band we needed to cater for. Just having this information in the datasheet made the component selection far quicker.
“Second, the Nordic chip, particularly when equipped with the RHCP antenna, has excellent receiver sensitivity. Signal strength is one thing when the satellite’s overhead, but quite another when it’s low in the sky,” Hibbett adds. “The extra decibels you gain from a particularly sensitive Bluetooth LE SoC make a lot of difference.”
“Third, Nordic’s DKs and software development tools are so comprehensive it makes the design process easier and therefore faster. For example, the SDK contains an Eddystone beacon example making that side of the process simple.
“Finally, although the Nordic stack [SoftDevice] is good, it is not accessible to external developers. So the fact the chip seamlessly runs alternative open-source stacks such as our choice, MyNewt, allowed us to modify the Bluetooth LE stack but still gain all the advantages of Nordic’s proven hardware, software architecture, and development tools,” says Hibbett.
Putting it into practice
Part of ESA Maker-Space’s requirement was for the company solving the engineering solutions to come up with application ideas. Firmwave’s suggestions ranged from the mundane — updating a fleet of beacons’ firmware saving a technician being dispatched or sending information to public installations such a bus shelters which could then be passed on via the beacon to passenger smartphones — to the life-saving. One example of the latter is getting emergency information to people where cellphone coverage is non-existent.
“Bluetooth hardware is ideal for these applications because it’s robust, inexpensive, has long battery-life and can be constantly updated while in-situ,” explains Hibbett.
“A lot of people carry cellphones to areas where there’s no signal,” he adds. “For example, imagine a mountain pass that has been blocked by a landslide. The satellite could be programmed to send information to beacons located at the bottom of the pass, notifying people that the road tens of kilometers up the mountain is blocked, improving safety.
“But even we can’t imagine many of the potential applications based on this technology. After all, who’d have guessed that Bluetooth LE — a technology originally designed for short- range communication between a smartphone and a close-by peripheral— will soon be beamed thousands of kilometers into space and then back down to Earth. Perhaps the Moon will be next.”