GNSS (Global Navigation Satellite System) constellations like GPS are widely used to provide Location Based Services (LBS) in automobiles and mobile phones. Now there are emerging opportunities â€“ and increasingly requirements â€“ to include GNSS in battery operated IoT devices, whether for asset tracking, determining driving patterns for insurance purposes or myriad other applications. Asset tracking alone is estimated by ABI Research to reach billions of things by the early to mid-2020s. Because of the extreme low-power constraints of these devices, and the extremely power-hungry nature of GNSS receivers, designing GNSS capability into a battery-operated IoT system is a daunting challenge.
The power consumed by GNSS receivers is due largely to the large amount of time they must spend time searching for and acquiring signals from several different satellites in order to fix a location. The signals they receive are often weak, since they must pass through the near-vacuum of space then through the various layers of the earthâ€™s atmosphere. The receiver needs time to process the signal so that it can rise above the level of noise, which takes about 30 seconds â€“ a time during which battery life drains from the device. Even in situations where the system is assisted in getting to first fix by LTE or Wi-Fi, research shows that six seconds are needed.
Given the increasing opportunity and requirement for location-based services in battery operated IoT devices, a different approach to GNSS receiver design was needed.
A GNSS receiver can be built in hardware or software or mix of both. Depending on the choice, there are tradeoffs in terms of flexibility and power consumption. The Ensigma design team saw an opportunity to add GNSS functionality to its communications IP portfolio and they recognized that the ideal solution would combine a software and hardware approach for a balance of power consumption and flexibility.
They knew that tackling the power consumption issue would be the main obstacle to designing a product that would fit within the power envelope of its customersâ€™ devices.
The team built the Ensigma Location GNSS IP on the foundation of the proven Ensigma connectivity engine, which incorporates an ultra-low power CPU core. They then employed many innovative techniques to build on the platformâ€™s efficiency, including software techniques and dedicated hardware blocks. But the team needed to do more.
The answer lay in the nature of the IoT devices themselves. Most traditional GNSS chipsets synchronize data in real time and are therefore continuously on, but not every IoT application requires an always-on, real-time approach. Many applications only need to deliver location data periodically â€“ think for example about finding a specific sheep in a field for annual medical procedures or shearing, pinpointing a streetlight that needs maintenance, or locating a stolen bicycle. Ideally the radios in these devices would only turn on periodically for extremely brief periods of time.
With this in mind, the Ensigma team devised a â€˜snapshotâ€™ approach, in which a short recording of the satellite signal is acquired as rapidly as possible, and then processed offline after the receiver powers down. Snapshot is designed to work with only a few milliseconds of a raw GNSS signal. In the Ensigma receiver implementation, the radio turns on for 100 or 200 milliseconds, captures the data at a very low sample rate, compresses the data and stores it in memory. To ensure accuracy the team the Ensigma Location GNSS IP with bit width flexibility, for example searching for satellites on 1-bit, and then tracking satellites on a higher bit rate. With the Ensigma snapshot technique, there is no need to track satellites or decode navigation messages in real time, leading to dramatically reduced receiver power consumption.
The snapshot technique, also known as â€˜capture and processâ€™, can reduce RF power usage by more than 99.5% in a GNSS receiver. This is a game changer. Typically, GNSS is restricted to devices such as smartphones and Satnavs that either use large batteries or can be recharged, limiting the opportunities and success of GNSS. Ensigma is enabling GNSS to be a viable option for other devices that are very low power and do not require charging for months such IoT devices, removing the barrier to mass adoption.
â€œSilicon and system vendors designing their own silicon need low-power, high-quality GNSS IP. Until now, theyâ€™ve had to either settle for a restricted set of features or use a companion chip to provide comprehensive functionality. ABI Research has consistently maintained the integration of fully featured GNSS with other wireless technologies has the ability to enhance the overall performance of the chip and lower the overall power consumption without compromising cost.â€ â€“ Malik Saadi, Managing Director and Vice President, Strategic Technologies, ABI Research.
â€œImagination has developed a product by looking at what IoT applications need, rather than pushing an existing design into an IoT role. The result is lower power consumption, and greater flexibility, which in turn enables the use of GNSS in applications where it would otherwise be impractical.â€ â€“ Bill Ray, Senior Director, Semiconductors and Electronics Research team, Gartner.
Building on the result:
Imaginationâ€™s Ensigma Location GNSS IP was introduced to the market in October 2018. By creating Ensigma Location GNSS IP, Ensigma now has a complete offering. Historically, Ensigma has been successful in both Wi-Fi and Bluetooth. By now offering GNSS IP, itâ€™s portfolio is complete and means Ensigma can offer a complete package for connectivity to its customers. For Imagination Technologies as a whole, this broadens the overall Ensigma IP product offering, which now provides connectivity, radio/TV and location.
Today, just three short months since its launch, more than 10 potential customers are evaluating Ensigma Location GNSS IP from around the world and we hope to announce a customer shortly.
Designers of GNSS receivers for battery-operated IoT devices are under many pressures â€“ the need to support the multiple broadcast frequencies needed for the different satellite constellations; to make antennas smaller, lighter and lower cost; to guard against increasing amounts of interference in the L-band RF spectrum; and to meet aggressive low-power consumption targets. The Ensigma team built a compelling IP solution that supports all major constellations, provides mitigation against RF interference, and enables customers to integrate the technology on their main system-on-chip to save costs versus using an expensive companion chip. They balanced flexibility and power by using a combined hardware/software solution. And importantly, they solved the power consumption challenge for battery-operated IoT devices by looking at the inherent nature of these devices and designing a solution specifically for their use cases. Through its use of cutting-edge snapshot technology, Ensigma Location GNSS IP can offer ultra-low-power location capabilities that usher in a new wave of use cases for the IoT. Simply put, it enables the creation of low-cost location devices that donâ€™t require the batteries to be changed, potentially for up to years at a time. For the IoT, it’s a game changer.
GNSS (Global Navigation Satellite System) constellations like GPS and others help us to find the easiest and fastest routes to our destinations, help us find our devices when they’re lost, and enable us to survey and map the pathways of our world. Until now, consumers have mainly used GNSS in their automobiles and mobile phones. Now there are emerging market opportunities for GNSS in the Internet of Things (IoT).
Within battery operated IoT devices such as remote IoT sensors and edge devices, wearables and health monitors, consumer mobile products, automotive after-sales products like insurance boxes and road tolling equipment, asset tracking devices, and others, Location Based Services (LBS) based on GNSS will have a major impact across our homes and cities. GNSS will be a key requirement going forward for IoT applications, many of which will mandate it for regulatory/insurance purposes.
Due to extreme power constraints, designing a GNSS receiver for a battery-operated IoT system is a daunting challenging compared to those in the systems we see in our vehicles and smartphones today. IoT edge devices can be untethered, needing to survive on either battery power or energy harvesting for weeks, months, or even years at a time without a battery re-charge or replacement. As such, the GNSS receiver inside these devices must have sub-milliwatt power consumption, versus the 20-30mW of traditional GNSS receivers. The industry needs a new approach to GNSS receiver design to make LBS a possibility for these new devices and applications.
Companies are looking to solve the power challenge in numerous ways, through hardware optimizations, software techniques, LTE assistance for improved time to first fix, new techniques to acquire location, and others.
In this paper, we will discuss design considerations and tradeoffs for a low-power GNSS receiver, including the main contributors to power consumption, and how you can design your receiver to provide the best solution possible within the power constraints of a battery-powered IoT device.
For the sake of this paper, we’ll focus on receivers for civilian bands used in consumer products (not military, maritime, emergency and other GNSS signal bands).
GNSS Receiver Basics
A typical GPS receiver system consists of a GNSS antenna, an RF front-end and a baseband signal processing engine. The antenna receives and amplifies the GNSS signals from the satellites, the RF front-end further amplifies the signal and digitizes it for processing, and then the baseband processor (also known as a correlator) decodes the digitized signals and performs a series of computationally intense signal processing operations to calculate position, velocity and time (PVT). The algorithms that convert the raw data into PVT information are part of the CPU, which can be either integrated with the baseband processor or operated separately.
Diagram 1: Typical Receiver Architecture
Traditional GNSS receivers acquire, track, and decode GNSS navigation data in real-time, and acquiring/tracking the signal is the most complex and compute-intensive part of the process. For an accurate positioning calculation, the receiver must collect data and correlate code from at least four different satellites.
The signals that are received are often weak, since they must pass through the near-vacuum of space, then continue on through the various layers of the atmosphere to the earth. This means the receiver needs time to process the signal so that it can rise above the level of noise. This takes about 30 seconds – a time during which battery life drains from the device. Even in situations where the system is assisted in getting to first fix by LTE or Wi-Fi, six seconds are needed.
Research by The University of Texas has shown that baseband processing accounts for over half of the energy consumed in modern GNSS receivers. Since power consumption generally increases significantly during the time that the receiver is searching for and acquiring satellites, there are many techniques designed to keep power consumption to a minimum. Most hardware GNSS receivers feature sleep or standby modes during which the RF front end and baseband processor are powered down, leaving only the CPU on. The CPU wakes upon receiving a signal then activates the RF and GNSS circuitry. There are also more aggressive designs where even the CPU is powered down.
The Constellation Matters
Today there are four major satellite constellations. These include GPS (originating in the United States), GLONASS (Russia), Galileo (Europe), and BeiDou (China). In addition, there are numerous regional Satellite Based Augmentation Systems (SBAS) systems that help improve accuracy, including Wide Area Augmentation System (WAAS) and the European Geostationary Navigation Overlay Service (EGNOS).
While each of the four major constellations is designed around a similar GNSS topology that is comprised of space satellites, control network and user equipment, each was architected in a different way and therefore building support for each constellation into the receiver has different power consumption implications.
Diagram 2: GNSS High-Level Architecture
There is one school of thought that GNSS consumes too much power in general to be used in battery operated IoT systems. To address this, companies are looking at other geolocation methods such as approaches for asset tracking using LWPANs.
But since the energy consumption of a GNSS receiver is often too high because it is tracking all four constellations in real-time, there are other approaches to controlling power consumption.
One possibility is to track only one or two of the constellations versus all four. While tracking all four will provide the most accurate solution (and most smartphone chips today support all four), not every application requires such pinpoint accuracy. If you’re designing an ultra-low power asset tracking device, it might not be necessary.
The two constellations which require the least power consumption in the baseband processor are GPS L1 and L2 C/A (for civilian use) and Glonass – which were the first two constellations in existence. While each of the constellations is designed to be ‘spread spectrum,’ the structure of the signals in the newer systems have different modulation techniques and signal parameters. Galileo and BeiDou were designed with larger signal bandwidth, more complex and longer codes, and higher navigation data rates. The aim is greater accuracy, but the changes have an impact on the receiver design since they require wider registers at higher operating frequencies.
Another challenge with tracking all four constellations is that some are on different frequencies and thus may require more than one radio. Only GPS and Galileo are on the same frequency.
Diagram 3: GPS Transmission Frequencies
Which constellation your receiver supports can also be dictated by more commercial concerns. For example, if your receiver will be integrated into a car that is imported to Russia, it’s mandated that the receiver support Glonass.
To ensure the most commercial and technical advantages, one option is to design support for all four constellations in hardware and make that support flexible to be programmed in software.
The sampling rate of a GNSS receiver, typically expressed in samples per second, or hertz (Hz), refers to the rate at which the analog signal is sampled in order to be converted into digital form. The sample rate determines how much processing is needed; in general, you can expect that 2x the sample rate leads to ~2x the processing. As you might expect, while a higher sampling rate leads to greater accuracy, it also leads to higher power consumption.
Which constellation(s) your receiver supports defines the sample rate. For example, the increased signal bandwidths of constellations like Galileo demand higher sampling frequencies. In keeping with their overall power consumption as outlined above, Glonass has the lowest sample rate and is the least complex to process, followed by GPS. Galileo is the most accurate, but its development has significantly increased the complexity of searching for a signal and tracking it—it’s 8x harder to find than GPS. BeiDou falls in the middle in terms of complexity.
The complexity of the baseband processor is impacted by the number of bits that are needed to represent the intermediate signals, the bit-width of the registers, and the minimum frequency needed to process a particular signal.
Each of the different constellations has different resource requirements in terms of the number of registers and combinational logic. As we’ve seen, newer constellations have increased complexity (e.g. increased code lengths and chipping rates), leading to the need for higher clock frequencies and additional memory. Keeping bit widths as small as possible can help to optimize memory and power consumption. As with other parameters, there is a tradeoff between power and accuracy.
Software or Hardware?
A GNSS receiver can be built in hardware or software or mix of both. A software-based GNSS receiver replaces the functions of the baseband processor with software running on a general-purpose CPU, and the general advantage is significantly increased flexibility. In addition, with a software based GNSS receiver, it’s possible to update the product with new features and functionality throughout the product life cycle and thus extend the product’s lifetime. This isn’t possible with a purely hardware-based receiver.
There is also an argument that power consumption can be decreased by designing the receiver in software because there is no need for a baseband processor. But with traditional software-based receiver technology, a significant amount of digitized GNSS signal data must be processed by the CPU which in turn requires a higher-performance CPU running at higher frequency, leading to increased power consumption.
To reduce power, software-based receivers often use Fast Fourier Transform (FFT) algorithms which sample a signal and divide it into its frequency components, but this technique leads to increased memory requirements and thus higher cost.
Hardware based receivers offer better efficiency in terms of both power consumption and computational load. This is because a hardware design can be highly optimized and specifically designed for the sole purpose of GNSS processing.
We believe that the ideal solution is one that mixes hardware and software to provide a balance of flexibility and power.
Not Every Hardware Approach is Equal
The need to track the major GNSS constellations as well as SBAS leads to a dramatic increase in the complexity of tracking and positioning algorithms and firmware. Tracking multiple frequencies and signal types on each constellation also leads to an increase in the number of channels that must be implemented on the hardware to track the different satellites. This leads to an increase in cost, power consumption and silicon area.
However, there is a way to design the receiver for multiple satellites without such a resource-hungry multi-channel-based approach. For example, Imagination’s Ensigma GNSS IP, which is a hybrid hardware/software receiver, is based on a dedicated processing engine which is used elsewhere for Wi-Fi, Bluetooth, TV and radio, and thus uses more generalized processing techniques. It is designed it to be flexible, and to be reused again and again – running the data through the correlator multiple times, and time-slicing it for the different satellites.
Minimize your Radio Time
As discussed, the radio consumes a great deal of power when it’s turned on while it is tracking and acquiring satellites. Therefore, it’s important to turn the radio off as quickly as possible, and to keep power consumption below 10mW – ideally 5mW.
A lot of GNSS chipsets synchronize data in real time and therefor are continuously on. But not every application requires an always-on, real-time approach. There are many applications which only need to periodically deliver location data. Examples include use cases such as finding livestock in a field, pinpointing a streetlight that needs maintenance, determining driving patterns for insurance purposes, locating a stolen bicycle, and many others. Even wearable fitness monitors don’t need to deliver location data continuously. For such applications, it is possible to periodically turn on the radio for extremely brief periods of time.
Take a Snapshot
The idea of a ‘snapshot’ approach, also known as capture and process, is to quickly turn on the receiver, capture the satellite data, store it in memory, and then process that data offline after the receiver is powered down. Snapshot technology is designed to work with only a few milliseconds of a raw GNSS signal.
Snapshot technology has already been implemented in some industry software solutions. Imagination has taken snapshot to the next level by implementing it in its Ensigma GNSS IP. In the Ensigma hardware/software receiver implementation, the radio turns on for 100 or 200 milliseconds, captures the data at a very low sample rate (generally 1-bit), compresses the data and stores it in memory.
Once the 128-bit memory is filled, the radio can be turned off and the Ensigma engine begins to cycle the data through the correlator several times in fast succession to get a fix on the satellites.
The processing time is not limited by the speed of the data transmission; it’s limited only by the clock speed of the correlator processing. With optimized bit widths, the Ensigma hardware doesn’t need to work as hard as other hardware-based receivers, so it consumes less power. One way in which Imagination has worked around the accuracy tradeoff is by enabling bit width flexibility, for example searching for satellites on 1-bit, and then tracking satellites on a higher bit rate.
Diagram 4: Hot Start Push to Fix Sequence
With the snapshot technique, there is no need to track satellites or decode navigation messages in real time, leading to dramatically reduced receiver power consumption. Since the GNSS receiver can be aggressively duty cycled, average power consumption can be further reduced.
Diagram 5: Duty Cycle Operation
Designers of GNSS receivers for
battery-operated IoT devices are under pressure to support the multiple
broadcast frequencies needed for the different constellations; to make antennas
smaller, lighter and lower cost; to guard against increasing amounts of
interference in the L-band RF spectrum; and to meet aggressive low-power
consumption targets. They must take into account the power consumption
implications of each element of their GNSS receiver from the constellation
choice to the baseband processor bit widths and memory sizes to the real-time
(or non-real-time) requirements of their application. For many emerging
battery-operated IoT devices, snapshot technology on a receiver that leverages
both hardware and software can provide an ideal ultra-low power GNSS receiver solution.
 Source: Radionavlab – https://radionavlab.ae.utexas.edu/images/stories/files/papers/lowenergy_ion2013.pdf
Appendix: The place additional information to which the judges MIGHT refer forunderstanding, but that will not be used for scoring.
Designed specifically for mobile battery-operated devices, the Ensigma GNSS IP supports GPS, GLONASS, Galileo and BeiDou as well as several satellite-based augmentation systems (SBAS) including WAAS and EGNOS.
Key features of the Ensigma GNSS IP include:
- Ultra-low power operation
- Ability to share location information with external wireless technologies such as LTE or Wi-Fi
- Options to share peripherals such as memory and system clock to enable reduced total system cost
- High sensitivity for indoor location
- Support for assistance information using external sources such as LTE and Wi-Fi to improve time to first fix
- Radio Frequency Interference (RFI) detection and mitigation
With Ensigma GNSS IP, customers can choose a configuration, from standalone to highly integrated, that works best for their specific system implementation. The IP can be configured for the lowest power or highest integration, and is designed to fit into any existing solution with the optimum level of design and resource re-use.
Article of interest:
Inside GNSS, 6th December 2018, What is Snapshot position and what advantages does it offer?
ABI Research Executive Foresight Report, November 2018
You can find out more here