[HH] More re: LoRaWAN and geolocation

[Stephen Ronan]

The Dutch telecom provider KPN completed a nationwide LoRaWAN network
early last summer. They're now rolling about geolocation services.
According to a user on the ThingsNetwork forums:
--------------
you can't get a LoRa KPN abo from KPN themselves, the provider
SIMpoint is delivering this for them. I got the following info from
SIMpoint:
LoRa with Geolocation: (costs per year)
Bundels Uplinks Downlinks Price
Often 36500 3650 € 15,30
Mostly 109500 900 € 17,75
I will probably need the 'often option', I have 36500 uplinks a year
with this option, that is 100 a day so more than needed.
https://www.thethingsnetwork.org/forum/t/looking-for-a-lorawan-board-only-geolocation/4419/9

========================

And, at least according to their public pronouncements thus far, the
ThingsNetwork via their first generation, soon-to-be-released
kickstarter gateways won't support geolocation based on high
resolution timestamping done at the gateways of messages from nodes.

However, as of this exchange on their forum last September:

Johan: "Also, we're going to support localization using WiFi BSSIDs,
also known as the MAC addresses of the access points nearby, directly
in the Handler for a seamless application developer experience. The
idea is that your devices briefly scan WiFi access points using cheap
WiFi chips (e.g. $5 ESP8266), packs it as LoRaWAN uplink message and
sends it to a special port for the Handler to pick it up, decode it
and lookup the location. This is not as low power as LoRa
triangulation, but consumes less power than GPS and is quite accurate
in urban areas (where LoRa triangulation is really bad still).
-----
hoonppark:  Regarding the localization using WiFi BSSIDs, I have a few
questions as follows:

What would be the accuracy? For example, it would be accurate within
20 meters or 50 meters range ?
Does it need just one WiFi BSSIDs, or at least three WiFi BSSIDs to work?
What happens if there is no WiFi access point near by?
Does the Handler calculate the location of the node? Or, does each
application need to calculate the location of the nodes?
According to the roadmap, it will be available in Q4 of 2016. Will it
be available by Dec. 31, 2016?
You mentioned "LoRa triangulation is really bad still". Do you mean it
is still bad even with the Semtech’s LoRa Geolocation Solution
announced on June 30, 20161?
Do you think this WiFi BSSID-based localization is better or
comparable to the publicly announced 'CSEM's LoRa-based GPS-free
Positioning Solution2'?

-----------
johanLeaderSep '16
@hoonppark; here are the answers:

That depends on various factors, see here
<https://en.wikipedia.org/wiki/Wi-Fi_positioning_system>. From our
testing so far using Google Localization API, we're close to 10-20
meters in the center of Amsterdam
More BSSIDs makes it more accurate, but costs payload. That's the
trade-off application developers need to make
Then you need to fallback on another way of localization. We are
sending the gateway's coordinates in the metadata
The Handler does that for you. Note that depending on the database
we're going to use, we may need an API key of even a subscription for
this
Yes
Yes
It's better in the sense that it will be more accurate in cities, but
not better in terms of power consumption. They positioning is just
standard LoRa triangulation; no extra magic
===========

Tangentially, another user posted a link to: EspWiFiTracker
https://revspace.nl/EspWifiTracker
  - Stephen Ronan