Use GPS and MEMS sensor combo for next-gen positioning and navigation

Editor’s note: This article was originally published in EE Times Europe.

Over the past decade, GPS positioning was transformed from a specialized high-end technology, mostly used by professionals, to a technology used by consumers on a daily basis. This transformation is the result of two generations of GPS-enabled devices that have commoditized GPS positioning in mass-market devices.  This article looks at the next exciting stage in GPS evolution that solves the problems of non-availability of GPS location information - often when the user needs it most.

The first generation of devices that appeared in the early years of the past decade, consisted of aftermarket personal navigation devices (PNDs) which enabled turn-by-turn navigation at a price point that facilitated mass-market penetration. The next stage of the revolution was marked by the appearance of the smartphone. Smartphones today offer various positioning capabilities, and heavily rely on the GPS functionality for accurate positioning and turn-by-turn navigation applications. GPS-enabled smartphones not only extend turn-by-turn navigation applications to pedestrians and cyclists as well as drivers, but offer positioning capability to any smartphone application, from local restaurant location to 'live' star maps and augmented reality. And beyond smartphones, low-cost positioning technology is appearing in other consumer products such as cameras for geotagging photographs. But the reach and scope of such applications could be significantly extended if the limitations of GPS performance in indoor and dense urban or forest environments could be overcome.

The appearance of the smartphone also led to changes not immediately visible to consumers. Two major trends are obvious when examining the shifts in the components that comprise modern handsets and smartphones. The first trend is the deployment of chips that converge several functions into a single integrated circuit (IC). Commonly referred to as "combo-chips", there are numerous combinations available today in the market including Wifi, Bluetooth and GPS.  The other trend is the massive use of MEMS based sensors.  Advances in MEMS technology in the past few years enabled implementation of multiple low-cost MEMS based sensors in the handset platform.  For example, many smartphones today typically contain three motion sensors--a MEMS accelerometer, electronic compass and MEMS gyroscope--which, when used together, provide a means of accurately tracking the linear and angular position, velocity and acceleration of the handset.  Until now, MEMS sensors were typically stand-alone chips (ICs), each performing a single function, but the future trend will be convergence and integration of motion sensors into a single package, together with sufficient processing capability to intelligently combine data from the individual sensors and provide meaningful motion vectors directly to the application.

The use of MEMS sensors enables mobile handset platforms, for the first time, to measure parameters that are related to the platform position and movement in the real world.  Such types of measurements are often referred to as “inertial measurements”.  The use of inertial measurements (or INS--inertial navigation systems) is well-known and established in the world of GPS and radio-navigation.  Inertial navigation is often used to improve radio-navigation or complement it in environments where radio navigation is hard or impossible. Examples for such situations in GPS receivers are dense-urban areas, underground or indoor locations.

Figure 1. Bringing together the capabilities of GPS and motion sensors deliver an improved user experience.

It is important to note the differences between the two methods of positioning. GPS or radio navigation relies on measuring the distance (or delay) to several known transmitters in order to triangulate the mobile receiver's position. Therefore, a GPS receiver directly calculates the mobile receiver’s position each time it performs a measurement. The downside of direct position measurement in radio-navigation receivers is that the measurement errors for a single measurement can be quite significant, even 10s of meters, and therefore receivers typically use dynamic estimation algorithms such as Kalman filters in order to reduce errors to acceptable levels.

On the other hand, inertial navigation system measure physical parameters of the platforms such as linear acceleration (accelerometers), angular acceleration (gyroscopes) and absolute direction of orientation (3D compass). By integrating over the acceleration data, a navigation system can calculate the position and orientation of the user, assuming the initial position is known. The upside of using inertial navigation systems is that the measurement errors are usually smaller than the errors resulting from direct positioning in radio-navigation receivers. However, since the receivers integrate over the measurement data, errors accumulate and lead to position "drift" when integrated over long periods of time. It is therefore evident that GPS and INS systems complement each other when combined.

In a typical combined system, the GPS receiver calculates the initial position, while the INS system is used to generate the position difference from the initial position. To avoid drift errors due to INS, the GPS receiver is used to correct the position periodically. Naturally there are optimized ways to achieve this combination and this is usually achieved with the use of an extended Kalman filter (EKF).

Currently, most GPS/INS implementations require the use of sensors that are aligned to the mobile platform and typically mounted in a fixed installation. A good example of this is the installation of inertial sensors in a fixed manner to a vehicle platform. The methods and algorithms used for GPS/INS navigation in such platforms do not usually perform well in situations when the INS unit is not aligned to the mobile platform, such as a person holding a cellular handset in his/her hand while walking. However, in recent years, and in anticipation of pocket-mobile positioning applications like pedestrian navigation and indoor navigation (in shopping malls, public buildings, museums, exhibitions, etc.), there has been a vast research effort on this topic with the goal of enabling constraints-free GPS/INS navigation. Recent results in several of these studies show promising directions in the commercialization of constraints-free GPS/INS navigation.

Combining the trend of converging multiple technologies into combo-chips, along with the requirement for both GPS positioning and MEMS sensors on mobile platforms, the next step in this revolution is to create combo-chips that contain both GPS technology and MEMS sensors. Other than introducing a higher degree of integration, platforms based on these chips would enjoy better navigation performance and enable for the first time true, highly accurate positioning in GPS challenged environments such as indoor or dense urban areas.

Recently, GPS receiver manufacturer teamed up with Baolab Microsystems, a European start-up company that developed a low-cost MEMS technology manufactured in standard CMOS. Now, combo-chips offer GPS and MEMS sensor solutions, integrating GPS and 3-dimensional electronic compass functions in a single chip product. This is the first step in enabling positioning and navigation in mobile consumer devices (mobile phones, cameras, etc.) based on GPS/MEMS combo-chips. The possibilities enabled by this technology are endless especially as this type of solution delivers significant cost reductions, enabling the overall solution to be made much less expensively than existing technologies.

Figure 2. Baolab Microsystems NanoEMS 3D digital compass, etched from the interconnect layers inside a standard CMOS chip

Figure 3. 'Under the hood' SEM view of the NanoEMS CMOS compass element

In conclusion, knowing exactly where you are is paradoxically often 'not available' when you need it most--in dense urban environments, when emerging from underground parking lots, in forested areas and increasingly for mobile devices, indoors (shopping malls, public buildings, etc.) as GPS cannot deliver in these locations. 

Ironically there is too much going on that blocks the signal and also confuses the user who cannot get visual directional clues.  By enabling high-quality dynamic navigation in these GPS-deprived environments using the augmentation of GPS with INS motion tracking data from MEMS sensors, consumers will be able to enjoy seamless operation of GPS-enhanced applications 'in their pocket' almost anywhere--especially when they need it most.

About the Author

Dvir Rosenfeld is VP of R&D at CellGuide Ltd.  In his role Dvir oversees the development of GNSS receivers and receiver components including RF, ASIC/VLSI, signal processing and software. Prior to joining CellGuide, Dvir held several R&D positions with the Israeli defense industry, focusing on signal processing and radar technology. Dvir holds a B.Sc. in Electronic Engineering from the Tel-Aviv University.