spirent.com

Use Case #2 for Accurate Everywhere Positioning: Sensors can fill in the "gap"

There’s no point in having a decent car, if you spend most of your time stuck in traffic.

The thought flashes through my mind as I drive through the streets of NYC, occasionally glimpsing snarling traffic blocking avenues and cross streets. The navigation app on my smartphone helps avoid all the major blocked points - using real time, crowd-sourced traffic data and location information from thousands of users, including the frustrated ones stuck in traffic.

The densely crowded cityscape makes the process a challenge. The app has to contend with real-time traffic data, as well as ensure that my position is calculated with a high degree of accuracy. One mistake – locating me in the wrong cross street, for example - could easily have me join the other dozens of annoyed, stridently-honking drivers.

In our last blog on this topic, Kunal ended with wondering how well his smartphone would have handled navigation within a city – and indeed, using a combination of GNSS and Wi-Fi AP fingerprinting, it handled it well enough to get me all the way, trouble free, to the exact same problem spot he encountered – the George Washington Bridge leading outside the city.

There are two ways to cross the bridge – the upper and lower decks. Both decks are open for cars to pass through, although the flow of traffic can be extremely asymmetrical. It’s hard to say which deck would be congested from merely looking at the entrance to each one (unless you are a New Yorker, which I am not). And so, I relied on my app for guidance – and indeed, the robotic feminine voice that had guided me so far confidently asked me to choose the lower deck. Emboldened by the skillful way she had guided me through midtown traffic, I drove into the lower deck (barely registering the cars with New York license plates cutting me off to get to the upper deck).

After a few minutes of smooth driving, I felt my heart sink as I realized I was stuck in a morass of flashing police lights, car exhaust fumes, and the echoing growls of multiple frustrated car engines, all enclosed within the tunnel-like walls of the lower deck. As I helplessly watched the cars speeding away to the upper deck on a nearby lane (thoroughly out of reach by now), I wondered in frustration – how could this happen? 

Glancing at my navigation app, I realized the uncertainty around my location had grown tremendously.  A lack of GNSS and WiFi – due to the enclosed nature of the tunnel - forced the app to resort to coarser methods to find my position. While I don’t know the exact reason for the choice the app made, I theorize that some combination of low-accuracy crowd-sourced data, and rapid traffic developments (an accident) had misled it.

How could this have been avoided? One way would be to have MEMS sensor based dead-reckoning kick in as soon as the device detected a drop in the other RF signals. When an accurate GPS fix is available (such as would be right before entering the tunnel), it can be combined with sensor measurements such as angular direction, heading, acceleration.  These can be used to track the user’s current position pretty accurately – for a short period of time. The bridge is not long to pass - even a few minutes of navigation with 5-10 m accuracy could have yielded enough data for the app to accurately differentiate between a traffic jam on the lower deck to potentially re-route drivers to the upper deck.

This example highlights the challenges involved in making hybrid positioning work.  Data needs to be shared across multiple positioning technologies and with a positioning server on the network. The device also needs to constantly monitor and measure the environment to select the proper mix of positioning technologies to use – while doing everything else the user needs (in this case, industrial rock by Rammstein helped deal with my frustration). Ideally, such scenarios could be developed in the lab – either by means of simulation or using field data – and devices could be benchmarked depending on their performance. Testing these scenarios in the lab would go a great way in ensuring customer satisfaction in the real world, as I was sorely tempted to throw my phone out the windows.

 

comments powered by Disqus
× Spirent.com uses cookies to enhance and streamline your experience. By continuing to browse our site, you are agreeing to the use of cookies.