Beacon investigation

The hottest topic in retail at present has to be iBeacons, a technology that is firmly associated with Apple who have invested a lot of time in studying the use of beacons incorporating Bluetooth 4.0, otherwise known as Bluetooth LE. This, they hope, will provide retailers with a method of targeting customers based upon their location. The question is does it work? And what are it's limitations?

Working within the retail industry, these are questions I have spent the last two months investigating, and in this post I will explain the technology, and reveal the outcome of these investigations, whose purpose was to provide a solution to location aware applications within a store environment.

To start with let me explain the perceived advantages of iBeacons. With the advent of Bluetooth LE (low energy) it is now possible to build a radio beacon that can run for years off of a single cell battery, and the very low cost of the technology means that these beacons are likely to cost no more than a Dollar or two, effectively making them disposable, at the end of two years you simply throw the old one away and fit a new one.

Secondly unlike older Bluetooth, you don't need to pair them with the device using it, the basic idea is that devices using these beacons only need to 'see' them, read the identification header that they broadcast, and measure the signal strength. Another beauty of the BLE standard is that these devices are much faster at booting up when a device talks to them, this is because when Nokia originally started working on Bluetooth 4.0 they heavily reworked the header the these devices broadcast, reducing dramatically it's size.

All of that looks good, and in most of the new generation of Bluetooth devices, it has worked faultlessly, and delivered all the benefits that were expected, but the idea of using these devices to locate a device indoors is very new, and there have been teething problems.

One of the biggest issues that this technology faces is that it's being promoted as a cure all for all of your indoor location ills, and because of this there are very few use cases that those working with the technology can agree to focusing on.

The only model that has had any real focus is the proximity one, a model that has been championed by Apple themselves. This is the approach most people have seen in use in the Apple stores, and works on the idea that if you approach a particular part of the store, offers and suggestions are displayed on your phone or tablet. On the whole this has worked after a fashion, but even this limited use case has been fraught with problems, and it's very much hit and miss if the technology is functioning when you visit one of their stores.

The model I've spent the last two months looking into, has been the location, one that many seen to be working on, but nobody seems to have cracked. Trilateration is the buzzword here, a similar approach used in satellite based geolocation, but this doesn't seem to scale down well to the much smaller distances we are dealing with with indoor location. To make matters worse the signal currently achieved from BLE beacons is not very stable, and although companies working in this sector of the market, are trying to come up with noise reduction algorithms to get over this problem, none seem to have cracked it.

Talking to the companies in the forefront of iBeacon development, they see the sky as the limit, everything from replacing NFC, and RFID, to geo-fenced payment systems, this is a sector of the market were there are a lot of people talking a good fight. 

Some use this technology as the reason Apple have yet to include NFC into their range of smart phones, and tablets. I guess only time will tell, either this will overtake and eventually replace NFC as the contact-less payment system of choice, or a future iPhone will incorporate NFC, which most will see as an indication that iBeacons failed to produce everything that has been promised.

As for my research, well for the time being we are calling this technology in its infancy, we will keep an eye on what happens with it, but for indoor location services, that we can feel sure will work on whatever platform we choose, we'll turn to Google's indoor location services instead.

Experiences with Google fuse location services.

Having tried for a prolonged period to get location via ibeacon working, with disappointing results (more about that in a future post), I've decided to give Google's fuse location service a try. Going down the beacon route wouldn't have been my first choice, but internal politics had, at the time, made using Google services all but impossible.

Before investing too much time on fuse location I have decided to test how effective it is in three already mapped locations, Waterloo railway station, John Lewis Oxford Street, and the British museum. The environment that I am developing for is probably closer to John Lewis, but testing a range of environments seems to be the best idea.

Location isn't the only issue I need to consider, platform is possibly more important, in particular the possibility of a system that can be used on a number of platforms, so for these tests I am using both an iPhone 5s and a Nexus 5.

The first environment tested in is Waterloo railway station, chosen in part because this is the station I come into every day. On arriving at the concourse I opened Google maps on the Nexus 5, and was immediately concerned and disappointed, the starting location that the Nexus gave me was at the far end of platform 1, almost the entire station away from where I was standing. Now I must disclose that the Nexus is the phone I currently use as my primary phone, and I know that my experience of mobile coverage using this phone at this location is often disappointing, so until I have tested this at all three locations I'll reserve judgement.

The iPhone however, bearing in mind that it was using the Nexus as a mobile hotspot, was very impressive, tracking my journey from one end of the concourse to the other with a consistent margin of only a couple of metres, and that margin was always the same bearing from where I stood. The location refresh was also very impressive, being almost real-time.

So in summary, ignoring the Nexus 5 for now, fuse location at Waterloo was very impressive, and definitely something that can be worked with. Next stop John Lewis Oxford Street.

Arriving at John Lewis and all concerns about the Nexus were blown away, using the in store WiFi on both phones, the position was tracked with a similar accuracy to that seen on the iPhone at Waterloo. Transitioning between floors was automatic for the Nexus, the iPhone needs you to select the floor your on, and apart from the third floor, I'm guessing the WiFi mapping might have been either in incomplete or out of date, neither phone skipped a beat.

In summary, at John Lewis, fuse location service behaved exactly as you'd want and expect it to, only gripe with the Nexus is that in Google maps the internal maps 'felt' zoomed in too much, an issue I don't expect to effect any apps written on the platform, using the fuse location service. Next stop the British museum.
Final stop on this fact finding mission, the British museum, and the one that I had been looking forward to, what a disappointment. Initially all looked as good as before, but as soon as you leave the central court yard the problems rapidly become apparent, art, antiquities and technology really do not mix.

Basically, for fuse location to have a fighting chance of giving an accurate position indoors it needs a decent Wi-Fi signal, and at the British museum Wi-Fi can be best described as patchy. Looking out for Wi-Fi access points, you fast realise that there don't appear to be any, I can only surmise that those running the museum did not want their exhibition spaces 'blighted' by unsightly tech, what a shame.

So to summarise, as an example of the fused location service, the British museum sucks. However in general across the three  environments, the service works incredibly well, and as far as my requirements are concerned this will be the solution I'll be using in the project I'm currently working on.