The three deadly sins of GPS

[Editor's note: We've already had one contributor from the mobile industry describe the opportunities that the new iPhone' built-in GPS service could provide to third-party mobile software developers. But GPS technology is still a work in progress. In the article below, Polaris Wireless Chief Executive Manlio Allegra, pictured, discusses issues with using GPS in dense, urban environments -- and solutions to the problem.]

With Apple’s announcement last week that the new 3G iPhone will be GPS-enabled, the media’s attention is once again on wireless location technology and the drive to improve performance for location-based services (LBS) applications. Once these capabilities are truly commercial, wireless carriers will be able to deliver timely, location-specific content to mobile users from real-time traffic reports and inventory tracking solutions to people finder and other concierge services.

However, GPS is not the technology that can get us there—at least not alone. Sure, GPS works wonders for most open sky applications like traffic navigation, but the technology fails to deliver high-accuracy locations in environments where end-users make most of the wireless calls—such as dense urban environments and indoors—due to obstructions from large buildings and walls. For this reason, the future of LBS cannot rely solely on GPS. Because the vast majority of today’s wireless usage is indoor and urban, GPS delivers a 50 percent solution (at best!), and here’s why.

1. Missing the L in LBS

Consistently pinpointing accurate location is essential for LBS, making sure consumers get relevant content in real-time. Most LBS applications that rely on GPS, Assisted-GPS (A-GPS) and their fall-back solution, Enhanced Cell-ID (E-CID), fail to reliably deliver high accuracy. These solutions can only produce urban and indoor accuracies within hundreds of meters, making the technologies undependable for delivering location-specific content city block to city block. Alternate solutions can determine location with a much higher degree of accuracy in challenging environments.

According to most research, more than 50 percent of cell phone calls are made in urban areas (cities) or indoors (office buildings, schools, malls, convention centers). Since the majority of wireless usage is concentrated in these dense settings, it is essential that wireless carriers use a location technology that is optimized for these environments. Unfortunately, GPS and A-GPS have serious limitations in these dense environments, because the satellite signals are too weak and corrupted by reflections from nearby buildings and walls. Location errors in these environments are more than 250 meters for many of the calls—errors that are larger than two city blocks.

2. Unacceptable Time to First Fix

Time is an essential ingredient in successfully providing LBS to consumers. Positioning latency—or Time to First Fix (TTFF)—determines how fast a service or content can be delivered to a mobile user. Time is of the essence because most users are typically on the move and need accurate, real-time and location-specific information. However, GPS can take up to a full minute or longer to make a first fix—an unacceptable delay—while non-GPS technologies can often determine a location in seconds. Customer satisfaction and the proliferation of LBS applications necessitates that information is provided in seconds. Consumers simply aren’t willing to wait minutes.

3. Throwing Satellites at the Problem

These severe GPS limitations in urban and indoor settings are not going to be solved in the foreseeable future. GPS solution providers are trying to address these problems by adding new satellite constellations (Galileo) to their receivers, hoping that density will yield better coverage. Unfortunately, they are finding that adding satellites only leads to more complexity and higher cost in the handset, without adding much indoor performance.

Satellites simply cannot provide the level of accuracy needed in those dense urban and indoor environments—no matter how many birds are flying overhead and how much power is added to the satellite signals. The fundamental problem is that GPS satellites fly at over 20,000 km from the ground, making the signals exceedingly weak. Satellites also have limited power budgets due to their solar power supplies. Even when the weak signals can be received, they are corrupted by reflections from nearby buildings and walls. And since GPS signals can never effectively penetrate deep inside buildings another solution is needed.

A More Reliable, Accurate Wireless Location Solution

Given these inherent drawbacks of GPS, Apple decided to enhance its GPS capabilities with a wireless location solution from Skyhook Wireless based on WiFi Access Points (AP). However, location technologies based on WiFi continue to lack complete coverage in dense environments, failing to cover ALL buildings in a wireless network—a capability that will not be added in the foreseeable future. How many times have you tried to log onto the Internet via a public WiFi network and couldn’t receive a strong signal? How many times have you gone into a building and couldn’t even get a WiFi signal? The coverage simply isn’t there, and that is an unreliability that LBS (or Public Safety) applications cannot afford.

In addition, the AP database on which the Skyhook technology relies is highly variable, as consumers and IT departments change AP locations all the time while lacking a reliable mechanism to keep track of the changes. The root of the problem is that WiFi is an unlicensed band that is unplanned, unmanaged and dynamic—essentially an ad hoc network. It is nearly impossible to produce a reliable and accurate location in a timely manner from the chaos—an absolute necessity for the proliferation of LBS.

Fortunately, there are carrier-grade solutions out there that are much better suited to LBS applications, including solutions that take a hybrid approach by leveraging multiple wireless location technologies to pinpoint highly accurate positions. These technologies range from hardware-based Uplink Time Difference of Arrival (U-TDOA) to Wireless Location Signatures (WLS). Polaris Wireless, Andrew Corp. and TruePosition are the leaders in this hybrid approach, building a combination of technologies that synergistically work together to accurately pinpoint the location of handsets. However, there are clear advantages between these competing technologies.

Andrew Corp.’s Geometrix Mobile Location Center (MLC) system uses a combination of U-TDOA, A-GPS and E-CID to determine handset location. TruePosition also incorporates multiple location technologies for its Position Determining Equipment (PDE) solutions, including a similar combination of U-TDOA, A-GPS and E-CID. These approaches with U-TDOA are inherently complex and costly to deploy and manage, with new radio hardware required at the base stations.

Taking what we have found to be a simpler and more cost-effective hybrid approach, Polaris Wireless has developed a reliable network-based solution based on Radio Frequency (RF) Pattern-Matching and A-GPS to determine handset location to a higher level of accuracy. The Polaris WLS Technology is a software-only, network-based solution that provides a much higher level of accuracy in dense and indoor environments than solutions that rely on other wireless location technologies or exclusively on A-GPS. Polaris is more reliable, commercially available, much easier to deploy and does not require new hardware or widespread buy-in from customers, making the initial deployments more scalable, cost-efficient and powerful. The technology has been successfully deployed in the US for public safety applications (E911) as well as for LBS applications.

This Polaris WLS technology is based on the principle that every location has a unique radio frequency (RF) signature. Like a fingerprint’s pattern of lines and swirls, a location can be identified by a unique set of values including measurements of neighboring cell signal strengths, time delay and other network parameters. Polaris capitalizes on this substantial suite of existing network measurement information, combining it with A-GPS to determine accurate locations to within 50 meters.

While GPS-enabled iPhones are certainly a step in the right direction, wireless carriers need to get past the hype surrounding A-GPS and look to more accurate and cost-efficient hybrid approaches that provides a rich user experience. Only then will the industry finally be able to truly capitalize on the power of location enabled applications.

Manlio Allegra is CEO of Polaris Wireless, a leading wireless location company that provides network-based software products for accurately determining the location of mobile phones.

Next Story: Report: For every iPhone 3G sold, Apple is getting $325-$425 from AT&T
Previous Story: Five billion iTunes songs sold. One looming problem: AmazonMP3 is better

Bookmark and Share

Tags:

Photo of Manlio Allegra

About the Author, Manlio Allegra

  • Logon Wheeler
    Nicely written article and very informative on the concept (and current limitations) of LBS.

    However since the article is written by the CEO of one of the technologies presented it isn't really a surprise that Polaris Wireless technology is cited as the most accurate an reliable of the alternative wireless location systems.
  • Great informative article. I'm not sure that fidelity/accuracy for a handset is that important though for most applications. I mainly use the existing location/map integrations in combination with search, not to navigate a car etc. I only need to know roughly where I am (zip code level would get me 80% there). ie. searching for "pizza", its ok if it shows me results within a few block radius.
  • It is a nice article which presents some state of art as well as problems of mobile LBS. In my understanding, the challenges to mobile LBS are more than the positioning technologies, though they are very critical to the LBS performance. For example, even the simple question how to make the user location data accessible to third parties is not that easy to be answered. Especially in a roaming scenario.

    Another interesting mobile LBS overview I can share is the tutorial presented in IEEE ICC 2008. It can be found at

    http://to.swang.googlepages.com/lbs
  • Interesting article. I was actually looking around the other day for something similar and come across an article which discussed why LBS may very well not be the future. No idea whether you are are interested in it.

    http://www.timdavis.com.au/2008/06/18/are-locat...
  • se419
    Polaris' solution to geolocation may be good, but at least some of the claims Manlio uses to deride his competition may be inaccurate and misleading.

    Regarding WiFi positioning and the service Skyhook provides, Manlio discounts coverage and accuracy based on the following statements:
    "How many times have you tried to log onto the Internet via a public WiFi network and couldn’t receive a strong signal? How many times have you gone into a building and couldn’t even get a WiFi signal?"

    My understanding of Skyhook's technology and service is that it is not limited to what you see when you try to connect your laptop to a hot spot. Manlio does not take into account that technologies used to sniff WiFi signals can divine much information from signals too weak for your laptop to deem relevant for display. They also use router mac addresses and don't need to rely on access point names for recognition.

    For a little more on how WiFi positioning works there's a video of Skyhook's Ryan Sarver at the following link:
    http://www.endpointenvironmental.com/W2MSN_meet...
  • Generally a good article aside from this being a white paper for Polaris. However, the importance of accuracy and precision depend on the application. For example, accuracy to a few meters may be irrelevant for metr0-targeted advertising and may be all that's necessary because there isn't ad inventory block-by-block. Sure, you could give someone a coupon for a store they are standing next to, but can you? Also there's a cost-benefit trade off for any infrastructure. Skyhook Wireless (mentioned above and in which I am an investor) has extremely low infrastructure costs because the public provides the beacons. At the other extreme, satellites can cost billions. Again, the investment depends on the application. (And yes, se419's comment is correct not only does one not need as strong a signal for Skyhook's technology to work but the device doesn't have to log on to the wifi network either....)
  • Interesting article and points on strengths and weaknesses with LBS. At our fund, we have found that from and advertising and transaction processing services perspective, the best results so far are using coupons to area locations with a call to action. Getting to within a few blocks is usually good enough, especially if you are tying to time of day and typical user activity based on location. It' s all about scaling volume, and the highest concentrations of users in densely populated areas are critical (urban home or work, and school). If the application utilizing a LBS functionality has a consumer facing Web presence, which is often the case, the consumer's registration information can close the gap to "back up" the several block area with their actual address and provide additional value as to their psychographic profile of being at home, work or school. This won't solve for a consumer "on the go", which is a whole other market and very much day-of-the-week dependent - consumers during the week spend most of their "purchasing" time in and around the same exact locations; weekends are much more on-the-go LBS driven without a back up of Web site registration to utilize.
  • Andrew Graham
    As an experienced LBS developer with aGPS application deployments onVerizon Wireless, Sprint, Alltel and others we are seeing faster initial acquisitions and associated consumer satisfaction than those quoted in the article (typically 5 - 20 secs).

    It's true that unfiltered aGPS readings suffer from signal reflection in urban settings and look like a scatter plot even in ideal conditions.

    However, we take MS based pings every 2-5 seconds to build accurate path traces (running, walking, cycling) with data conditioning that produces results within +/- 2% of actual. Here's an example from the NY Marathon.

    http://bimactive.com/ba/ui/route_detail.php?f_r...

    The biggest issue I see is with building penetration where hybrid approaches may help. References to aGPS being "overhyped" are a bit of a mischaracterization based on our experience with accuracy sensitive applications and the user experience of subscribing consumers.

    Andy
  • Perhaps the location battle is already over.

    See http://www.andrewgrill.com/blog/index.php/2008/...

    where I pose the question Has Google won the location battle already?
  • shrirang
    Great Article by Manlio on LBS. I guess the important part is to commercialize it, esp in areas where there is no mandate by the government to have Location determined.

    Local Coupons probably is very well suited for commercialization. Buddy finder has had a hype around it, but will have to wait and see. So cellfire and polaris solutions together can become a good combo.
  • Anon imous
    The Polaris Wireless solution has several drawbacks including poor accuracy which can only be as good as the database which in turn depends on continuous data collection and calibration and also has problems with locating power controlled CDMA based terminals. In short, the accuracy will never come close to what AGPS can provide. GPS companies are increasingly adopting INS and DR techniques for uninterrupted location in urban areas. The 50m claim here is simply what you would expect from the CEO based on performance at select testbeds. In reality, the accuracy is of the order of several hundred meters and cannot be relied on for navigation purposes. Indoor positioning has been talked about for several years now and would be a nightmare to implement given the need to create a database that includes signal strength vectors from every floor at extremely small grid spacing. In short, WLS is good for small carriers that are looking for cheap software based solutions but can never hope to upsatge or complement GPS and AGPS solutions.
  • xiaodi
    I completely agree with Anon imous. The 50m accuracy is the theoritical best the WLS model can achieve based on the statistical nature of the model and the PSD database (20m by 20m grid resolution). It maybe cheap to deploy (pure software based), but it is very expensive to maintain. You need to constantly remeasure and calibrate the PSD database. Based on my experience of the WLS, it accuracy is around 200m- 1000m.
  • Marty Feuerstein
    The previous post from "Anon imous" brings to mind the fact that every location technology has pros and cons. That's why hybrid approaches combining different technologies are so powerful. GPS and A-GPS are wonderful technologies and in open-sky conditions, their accuracy is hard to beat. But for all practical purposes GPS/A-GPS doesn't work indoors, where much of the wireless usage is concentrated these days. That creates the need for complementary technologies to drive location coverage indoors and into urban canyons. In those challenging areas, other technologies can and do perform better than GPS/A-GPS --- those are exactly the areas where network-based approaches like Polaris's work best because the cellular network base station densities are high. Inertial Navigation Systems (INS) and Dead Reckoning (DR) are technologies that can be used in high-end devices or in-vehicle navigation systems, but the cost, size and battery drain prohibit their mass adoption on cell phones, and more importantly they don't solve the indoor problem where even $1000+ INS/DR systems don't perform well.

    To correct some misunderstandings in that post, the Polaris system does work for locating CDMA terminals, whether they are using UMTS/WCDMA or CDMA2000. The CDMA measurements used for location are from pilot control channels, which are not power controlled (they are broadcast at constant transmit power). The claimed 50 meter accuracy at 67th percentile for urban environments comes from multiple independent blind tests conducted by large wireless service providers in places like New York, San Francisco, Tokyo, London and Toronto.
  • Ciao Manlio, presumo tu sia mio cugino, io sono Massimo figlio di zio Piero.
    Se sei realmente tu mi farebbe veramente piacere scambiare due chiacchiere, anche se solo via email.
    Un abbraccio
    Massimo Allegra