Posted by Kipp
Jones

Read All Posts

This post originally appeared in IoT Agenda. This is the third in a three-part series on IoT, location and TDOA. Read part one and part two

As discussed in my previous posts, the network-based localization provided by your low-power wide area network operator may not meet your use case requirements for accuracy. In those cases, adding Wi-Fi into the mix can provide additional accuracy and flexibility.

There are multiple ways to use Wi-Fi in your localization system when your device uses an LPWAN:

  • Sigfox networks can provide a Wi-Fi localization gateway to add location accuracy
  • LoRa networks have two options:
    • They can make use of both time difference of arrival (TDOA) and Wi-Fi to provide a hybrid localization solution
    • You can use the LoRaWAN network to request a Wi-Fi location from a service provider
  • LTE-based LPWANs allow you to pass a Wi-Fi location request directly to a service provider

An example hybrid LPWAN-Wi-Fi localization can be witnessed in a recent test we did to demonstrate the power of hybrid Wi-Fi location on a LoRaWAN deployment.

LoRaWAN networks make use of TDOA to provide a low-power localization capability. But the limited accuracy will also limit the use cases for which this proves sufficiently accurate. For use cases that demand higher accuracy, it is often preferable to pair LPWAN with Wi-Fi for optimal localization. This can provide the best of both worlds, offering universal coverage and high accuracy.

Certain environments, such urban and dense urban morphologies or indoor scenarios, are highly susceptible to significant multipath and attenuation of radio frequency signals. Although multipath can be mitigated through complex algorithms, there are opportunities to improve on TDOA-based locations. One simple and cost-effective method is to use a hybrid system by incorporating a complementary location technology, such as Wi-Fi.

Skyhook combines our decades of TDOA experience with nearly two decades of innovation and commercial deployments of Wi-Fi positioning capabilities around the globe. By adding a simple Wi-Fi scan into the payload via an LPWAN, you can significantly increase the accuracy of your device location while having minimal impact on both battery and network consumption.

This can be used independent of a TDOA system or in concert with TDOA to provide a hybridized localization that takes advantage of both signal sources.

In a recent large-scale test of ~27,000 test points in San Francisco, we demonstrated 21-meter accuracy at the 60th percentile using this technique. The figure below shows these results.

Figure: Large-scale Wi-Fi localization test in San Francisco. Source: Skyhook

This same power of Wi-Fi location accuracy can be achieved over other LPWANs with minimal battery and network consumption. A simple Wi-Fi scan on the device and a single network request can be comprised of under 100 bytes with no loss in accuracy; fewer bytes can be used, but it may cause a reduction in accuracy. These localization services are available as cloud services and are often available with different pricing models to fit specific business requirements.

In conclusion, LPWANs provide much-needed capabilities to enable many of the evolving use cases for IoT at scale by lowering the hardware and network costs for deploying IoT devices.

It is certainly true that coarse location can fulfill the requirements for a number of use cases, but to be truly successful for high-accuracy localization requirements, you may need to look outside of the networks themselves to help solve this need. One possible solution is to combine what is provided by the network operator’s network-based localization along with higher-accuracy technologies, such as Wi-Fi.

No matter what you choose, you want to make sure that the system matches your localization needs along a number of axes:

  1. Accuracy at various percentiles. Don’t be satisfied with a single-number accuracy measure, very rarely will that provide the complete picture.
  2. Accuracy in different regions/morphologies.
  3. Availability and yield — how often is a localization attempt successful? How often do they fail?
  4. Power consumption.
  5. Network bandwidth consumption.

It is much easier to take the time to make sure you get this right the first time than it is to go back and re-engineer a V2.0 solution that meets your needs.

Topics: Iot location Data Analysis Location Intelligence lpwan

   

Comments