Estimated accuracy of cell tower based location services with NB-IoT or Cat-M1?

Hi!
I have a fleet of narrowband and cat-m devices around the world. I want to determine their rough position by using the cell tower information. I tried CellLocate from uBlox and a few different SIM vendors also try to provide location maps. None have been to satisfaction. CellLocate gave me a plus-minus of 800 km (this was perhaps a year ago)... Back then I heard that I should not expect this to work as well on the NB network. I wonder if anyone can share experiences from Nordics cloud services, specifically through the NB or Cat-M networks. Is there a map or anything of the sort that allows me to estimate accuracy in different locations?

My devices (52850+uBlox  modem) are stationary and I would be ecstatic with something like 500m accuracy, but even several km accuracy would be very helpful.

I'm also designing a similar future product around the 9160.

Parents
  • Hello, 

    Our nRF Cloud location services can provide <700m accuracy for multi-cell and <1000m for single-cell according to the nRF Cloud documentation for Location Services, but please note Accuracy values are for reference only and not guaranteed.

    Kind regards,
    Øyvind

  • Yes I saw those numbers and the "which neighbourhood" and "which city" infographic.

    What I was looking for was a slightly more nuanced description. Are those numbers typical with NB as well? In/around almost any big city? How often/in what kind of locations does it simply not work at all? In Europe? America as well? What about the middle east?

  • The cellular positioning is based on what cell tower the device find around itself, this means:

    In the case of LTE positioning, these estimates depend on whether the device can find multiple cells (MCELL). In rural areas, the device might only find one or two cells covering a 10+ km radius. Cell-based location assistance should not be measured against the accuracy of smart phones that can use Wi-Fi location assistance or GNSS receivers and processors.

    From what I understand of the cellular positioning technology LTE-M is probably better for cellular positioning due to the cell IDs being the as with legacy LTE, while NB-IoT uses different cell IDs.  There's a lot less data available for NB-IoT cell IDs and these cells are bigger, making mapping more challenging/less accurate.

    Based on this, the devices sends the information of the nearby cells to nRF Cloud and then location is calculated by the cloud service and sent back to the device. nRF Cloud cell database should cover most of the world. 

Reply
  • The cellular positioning is based on what cell tower the device find around itself, this means:

    In the case of LTE positioning, these estimates depend on whether the device can find multiple cells (MCELL). In rural areas, the device might only find one or two cells covering a 10+ km radius. Cell-based location assistance should not be measured against the accuracy of smart phones that can use Wi-Fi location assistance or GNSS receivers and processors.

    From what I understand of the cellular positioning technology LTE-M is probably better for cellular positioning due to the cell IDs being the as with legacy LTE, while NB-IoT uses different cell IDs.  There's a lot less data available for NB-IoT cell IDs and these cells are bigger, making mapping more challenging/less accurate.

    Based on this, the devices sends the information of the nearby cells to nRF Cloud and then location is calculated by the cloud service and sent back to the device. nRF Cloud cell database should cover most of the world. 

Children
  • Thank you!

    due to the cell IDs being the as with legacy LTE

    Ok! To me that sounds pretty much like if you have communication you'll at least get accuracy on the scale of the coverage of one tower, since the position of LTE cell towers are quite thoroughly mapped. Great!

    less data available for NB-IoT cell IDs

    Yes this is the tricky part isn't it. Do you have any quantative information re. how sparse the data is? Or how quickly the data is (isn't`) growing? In different locations?


    nRF Cloud cell database should cover most of the world. 

    Even for NB? This seems to contradict your comment about less data. My previous experience with competing solutions indicates that the locations around Europe where I tried it were not mapped, at all.

  • Hi AlexanderOhm, 

    I noticed you had a few open questions still: 

    Yes this is the tricky part isn't it. Do you have any quantative information re. how sparse the data is? Or how quickly the data is (isn't`) growing? In different locations?

    Unfortunately we do not have hard numbers on this. I can say that our upstream vendors are constantly expanding their databases every day. So if you cannot get your request resolved today, we could acquire the data later on. 

     

    Even for NB? This seems to contradict your comment about less data. My previous experience with competing solutions indicates that the locations around Europe where I tried it were not mapped, at all.

    NB IoT cells certainly have less coverage than others since most devices do not use them directly, making it harder for our vendors to collect their locations. You could try a service like OpenCellId to approximate the location. NB IoT cells are more sparsely placed than others, so accuracy on these cells is going to be measured in KM usually. 

    Best,

    Cole

Related