IoT Symposium 2016: Indoor Positioning and Tracking



well good afternoon first of all I'd like to apologize for the sudden change erricka supposed to be presenting today but unfortunately she is feeling unwell so I'm kind of presenting on her behalf and today I'd like to switch gear a bit we'll talk about a little bit about the implementation instead of high-level pictures so mi-24 there you go so this is the agenda first I'll start with a company introduction and then we'll talk a little bit about the applications of indoor tracking using Wi-Fi and then we'll talk a little bit about the challenges that we've encountered so far with inter indoor positioning and finally talk a little bit about how this all fits within the IOT framework as a company we're local Hong Kong based company our focus is in the implementation of indoor tracking but indoor tracking is basically an infrastructure technology you can take indoor tracking and apply it and many technologies and we as a start-up company we don't build the end-to-end solutions but instead we work with partner companies who employ Taylor who integrate our solutions and our technology and then tailor it for specific applications and today we'd like to talk a little bit about what we've done in a healthcare environment throughout our experiment throughout our work within different companies and different projects and we've seen quite a few challenges in your positioning and then we solve them to experimentation research and development and some of them will go through today and so that you get a feel of what kind of challenges so that we get in the indoor positioning first of all about technology overview when we talk about when we think about positionings we automatically think about GPS well GPS is a technology that was created specifically for a purpose of positioning we all know that GPS is based on satellites in the sky it works very well when you're outdoor but when your indoor you cannot get satellite signals and so GPS doesn't work very well there are a lot of invert the tracking technology that's being introduced in the past we spoke a little bit about several of them this morning for example RFID ibeacon that is based on Bluetooth Low Energy and then we also have ZigBee that is becoming more and more popular but when we apply indoor positioning to some healthcare applications one of the key requirements is that it has to work anywhere it has to work in outdoor it has to work in your flat it has to work in MTR it has to work basically everywhere right and when we require a technology that's ubiquitous as that and basically it narrows you down to two technologies which is cell cell tower mobile tower and secondly is Wi-Fi because these are the only true ubiquitous technologies for any indoor tracking in our work we've been focusing on Wi-Fi simply because I'm comparing a Wi-Fi with a mobile tower Wi-Fi seems to be more accurate in our experiment so right at this point right we we know one thing outdoor tracking is based on GPS and a technology that was built specifically for positioning however for indoor tracking we rely on telecommunication infrastructure so when we compare the two different systems two different architectures the which has to be completely different indoor tracking is generally based on what we call real-time location system or our TLS it follows the IOT infrastructure is actually quite nicely for the tracking devices we have sensors in the picture over here there we have the red yellow and blue tracking devices these are simple sensors if these are Wi-Fi tracking devices all they do is that they collect signal strength from the surrounding Wi-Fi access point and then they turn around and they send it to the location server the location server is were doing most of the intelligence is it has a knowledge of the entire devices that we deploy out there it also has knowledge of the floorplan it also has knowledge of the fingerprint map and also the locations of each access point within the floorplan right and consistent with the Internet of Things technology we can supply the collected data to an analytic server so you can analyze the data in so many different ways you can tell like distracting particular tracking device how much time it spends at which locations you can tell like what kind of alerts that's been raised by each tracking device and what kind of battery level history that you have for each device all that that integration is possible and finally there is a user interface tracking device basically the old function of the user interface is to determine where the D on tracking devices are right so a communication protocol between the tracking device and location server is the m2m communis the m2m communication model so as you can see basically the indoor location technology follows the IOT architectural quite nicely and that's why we've been able to integrate that into an eye-to-eye light with solutions quite easily so some of the applications of indoor tracking using Wi-Fi in healthcare the most obvious one is we're tracking the elderly and children in this morning here we heard the term that was use which is healthy aging so that's the kind of applications so for this as I mentioned it's an application that has to work everywhere especially in a place like Hong Kong where people spend a lot of times underground are indoor right we need a technology that works outside and also inside we typically use GPS route or tracking and all the Wi-Fi for indoor and outdoor so the question here is that why do we need Wi-Fi for outdoor if we already have GPS it's because even when you're outside any u.s. GPS it has to have visibility to at least three satellites unless you have visibility to three satellites GPS doesn't work right so if you're sitting in the middle of two tall buildings having a clear view of sky doesn't work but in those cases we can still leverage GPS there sorry we can still leverage Wi-Fi there are two ways that you can use Wi-Fi in this kind of environment the first approach is basically to leverage existing location services and there's quite many Wi-Fi database that already exists some of them are location Google location service MLS they have millions and millions of access points database already in their network one database that I know they claim to have 900 million access points in the database already so we're not planning to reinvent the wheel like we leveraged whatever service that's already available out there and in a lot of cases especially in a public area such as Airport and shopping mall the accuracy is pretty good because they actually surveyed those areas right and the example over here is that we have this little girl who's inside festival lock and Google location service is accurate enough within that environment right but if you have used Google location services at all you'll realize that the accuracy when you're inside private area becomes less when you're in your own flat or when you're in your office for example Google location service will show that you're outside the building and not inside the building that kind of accuracy sometimes is not sufficient because if you live in a high-rise or you're inside the high-rise train you could be anywhere within that high-rise and to locate a person given that information that you're outside may not be sufficient from time to time for that basically we apply something that's called customized tagging so basically when you know that you're a certain location for example in your own flat we have the ability to pack that locations and basically what it does is that it associates the Wi-Fi access points that's visible from that locations to that location so the next time you or any device that you own enter that location it associates that with that particular location just by looking at Wi-Fi fingerprint right and this information is private so basically if you tagged a location to be your bedroom that is not visible to anybody else who uses service that's only useful to you you probably don't want everybody to know where your bedroom is so that's basically the technology that we use right in terms of tracking device somebody mentioned this morning the smartphone is actually a very good tracking device and we completely agree for this particular applications smartphone is preferred compared to a dedicated tracking device for example I have 11-year old daughter and her mother really wants to know where she is all the time and she never forgets her mobile phone so but if we give her a GPS tracker that she hangs on her belt I'll get you out guarantee you she'll never carry that it's the same as my father who's 80 years old he's still driving which scares all of us so we care about where he is and whether or not he's ok for even for him who's 80 years old he doesn't forget his art phone but if we give him a tracking device he will definitely not carry that around right so for this type of application we found that I'm having a smartphone app that runs on their smart phone is actually the best tracking device and a lot of nice thing about him using smartphone s tracking device is that you can integrate that easily with Bluetooth wearable devices maybe you want to have a bluetooth with some bracelet where you have a button that you can press for SOS that can monitor heart rate for example that erases an alarm if it detects a stop and heartbeat that's also possible with smartphone applications in a map here we show that there are two there are two maps there are both based on google map we offer selectable map we can leverage the google map but in some cases and we do want customized for plans so you may ask like why do we need customized for plan if you already have Google map it's because there are a lot of areas especially private areas that are not covered with Google map for example you're flat right Google map doesn't have the floor plan and exact location of your of your flat and even office buildings Google Maps and have that for public locations like shopping malls they have a pretty good map but this is why we have custom floor plan for some private areas basically and the second reason that's quite important is the privacy of tracking information some organization and some people just don't want to send their locations to Google because it's a public service and they have their own privacy policy but they want to confine all their tracking informations to just within their own system right so that's reason why some people still prefer to have their own floor plan and we support that too as what we see in this application over here the second healthcare-related applications for indoor tracking is what we call the staff and visitors safety and locations so some examples over here is that health care workers in unsafe environment nurses work at night in the mental hospitals quite the dangerous kind of job so we offer tracking here as you can see over there we use the hospital word for plan instead of Google map and that works too and another application is that you can track visitors to hospital for example so that you know if they wander off to a certain restricted area sensitive areas that they're not supposed to go to in u.s. they also use this particular technology for school teachers in the case of disaster of violence it's in case the teacher process an SOS button they will notify the local police who will come directly and provide assistance basically now if you think about right this is more of an enterprise kind of applications than residential applications and for this type of enterprise application what we found is that it's more practical to have a dedicated device like for example in this case here we have a Wi-Fi badge so basically it's a badge that collects Wi-Fi information with a Wi-Fi sensor and sends that information the location server that performs the same kind of thing is smartphone app the Wi-Fi back here is equipped with a SOS alert button and when the SOS alert button is pressed it'll automatically notify security who knows exactly where the button has been pressed and can send assist there and there's a display here that you can use also to broadcast messages so in case there's like you can one you want to broadcast weather reports or emergency situation you can do that too and finally you can impose a geofence around for a to implement area restriction even in indoor environment Wi-Fi badge next I'd like to talk a little with the challenges some of the problems that we've seen when we implemented this kind of solution as well as what we found the solutions to be the first is that the one of the questions that I keep on getting is that Google+ already has some sort of location sharing services why can't we use Google+ for tracking instead of building a whole new system the short answer is the shown in the bottom basically there is a difference between location sharing and location monitoring and location monitoring or what we call occasion tracking requires greater trust and what do I mean by that first let's define location sharing versus location monitoring the picture of location sharing is shown there in the middle the red phone there and basically it broadcasts locations to the yellow phone and the blue phone so that the yellow phone knows where the red phone is and the blue phone also knows where red phone is and the other two phones are basically doing the same thing so they kind of know where each other is right and this is what you can get from Google+ if you will it follows kind of the peer-to-peer model where as what how do we define location monitoring is that we have tracking devices the picture on the left there shows the red tracking device a yellow and the blue one all sending the information to a common server and from the UI there you want the person who's monitoring the service and can see where all the tracking devices are so the scenario there an example is like tracking the elderly like when I'm my grandmother I know where her location is but she doesn't know me to know where I am so if you will it's more similar to a client and server kind of application right so the major difference between the two is who has the information control in location sharing the information is being controlled by the person who is carrying the device who's being who's carrying the phone they can shut it off if they want or they can share it if they want right whereas in the case of location monitoring the person who is doing the monitoring has control of the information this is important because when you think about the kind of information that we need for location tracking services there are a lot of information there are just not shared for example location history when you do sharing you share the current locations but you never share where you've been in the past 24 hours with your friend and that information is considered too private and the second is battery level when you have a phone and you're sharing your location using a mobile phone you don't send with your battery and back the battery level however as mentioned earlier battery level is actually very important things that you monitor when you're doing tracking because if the battery goes out then you lose visibility of this thing so what you want is that you want the system when the battery level goes below certain level it will sent an alert so we can tell the person who's carrying it to recharge the tracking device or so on right so what we're saying is then location sharing services all these information are not sure not because they are technically impossible to do but because there is no such trust relationship between a person who is caring locating the tracking device and a person who's monitoring it and this goes back to the privacy policy and this is one of the challenges that we face when we try to use Google+ sharing location sharing services for location monitoring and that's the reason why it's necessary for us to develop a separate system and to perform only for the sole purpose of location monitoring the second question that I normally get is that why do we need to develop a separate tracking device when we can get the cheap Android phone and the answer is it has to do with the accuracy of smartphone when you think about it what is the ideal tracking device RTLS that is based on signal strength relies on one single path matically relationship which is shown over here which is the signal strength is inversely proportional to the square of distance what this means is that if I am standing at one place and I know what the signal string of an AP is at that location at this location right I can calculate the distance between the tracking device and the access point if I can see three access point from where I'm standing then I can then I know the distance between the tracking device and the three different access points I can use triangulations to determine the locations of this tracking device so this formula here is the whole basis for indoor tracking using RTLS that's based on signal strength by the same token a technology that is based on this should only depend on the signals on the distance what I'm trying to say is that if this is your tracking device and I turn it like that right it should give me the same signal strength reading because I'm standing as I'm location I have moved I want to see the same signal there is a very economical cheap and typical Wi-Fi antenna that we that has this property which is the only directional antenna that we've shown over there as you can see from just by the symmetry of this antenna no matter which way you turn it you're going to get the same signal reading so this is the ideal antenna for indoor location training and talking when you look at smartphone Wi-Fi antenna however the picture on the bottom right there this is the Wi-Fi antenna of iPhone 6 it's easy for you to see that how this antenna would give different reading as you rotate that around so the accuracy of this device would be impacted if you don't orient it the same way as when you took the previous reading on top of that different smartphones actually have different antenna designs this is just the antenna design of an iPhone 6 a samsung phone different type of samsung phone would have different Wi-Fi antenna design and on top of that further even the world based on our experience if you take two mobile phones the same brand same model a lot of times they have different signal readings as well so basically the answer is that the challenge that we face in using smartphone as a tracking device has to do with accuracy if we want more accurate indoor tracking capability than we am resort to a tracking device that we have better control over the antenna as well as the components and therefore we can make the signal strength more uniform the third challenge that we face is the what we call the access point solution this is probably one of the most important challenge that we need to that we need to take care of what do I mean by that when I'm sitting standing over here and what I was sitting at my seat right I looked at the Wi-Fi that's the fact around this area and I detected 15 Wi-Fi access points so the question is for indoor tracking which Wi-Fi access point do I use do I use all 15 of them or do I use only 3 and if I use only 3 which three do I use right first of all there la there are a lot of mobile access points the first thing that we do when we get a list of access point is that we'll get rid of our mobile access points as I mentioned earlier the whole premise of indoor tracking is based on the relationship that signal strength is relation is report inversely proportional to the square of distance by definition mobile access points they move around they come and go and if we use them for say for our calculations it will completely screw up our calculations so that's the first thing that we do the second is what we're seeing is that there's more and more beamforming AP beamforming AP is them I see them more and more they're typically use in Stadium and highly densely populated areas they have very narrow beam to focus their transmission only in one segment one direction basically while the actual relationship between distance and strength still has to obey the laws of physics and and therefore the mathematical relationship the problem with performing AP is that it is not uniform or round it's more like a directional right so it makes trilateration concept more difficult to implement so that's why for beam forming in a places where we have mode of beam forming access points we typically use fingerprinting and supposed to trilateration x' and oops I think I move too fast and finally the challenge another challenge that we need to deal with is the temporary variation in signals what I'm sitting over here somebody walks in front of me it completely disturbed signals because of interference reflection attenuations Y AP instability and so on so it's important that we perform the in AP selection intelligently and we do it based on a sample per sample basis for every sample that we take we determine which access point that we use and my last slide here as you can see that our implementations implement and machine-to-machine and IOT framework among the devices system and services the tracking devices would send infra tracking environmental variables directly to the to the location server without any user intervention and the location server would send back and configuration information directly to tracking device so basically the machines talk to each other I'm told that I'm my time is up so I'm going to finish this so additional information can also be integrated like body temperature and so on and the additional information that we collect that can be uploaded to an analytic server so that concludes my presentation I thank you for your attention thank you mr. Phan are there any questions for mr. fan today yes yes the main choir several space issues its absolute and yet you take a lot of the calibration before yes absolutely if the area coverage is very large then how are you going to do it we typically if the area is large and it's a compromise between the accuracy and also the the time that you want to spend you're absolutely right that there is definitely one thing that we have to pay attention for fingerprinting but when we consider the alternative that's related to its high latter Asians it has its own challenges to that we have to know exactly the location of the access points for the purpose of health care and monitoring residential in from locations we don't always have access to information on a precise location of the access points so maybe fingerprinting is the only options in those cases okay thank you thank you

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *