Thursday, November 2, 2023

Various Anchor-Tag Configurations in UWB

 In UWB literature, there is a notion of tag devices being mobile, lightweight, energy constrained UWB devices that typically users or robots would carry. Anchors on the other hand are UWB devices that are fixed in predetermined locations and serve as reference points to perform relative localization. Research at my lab has explored many relationships between tags and anchors and has challenged the notions of what is "fixed" and what is moving. In this blog, I will run over some of these configurations and point you to papers from my research as examples.

Multiple fixed anchors, infinite number of tags, privacy preserving:

If you are looking for an infinitely scalable indoor localization solution, you should read the PnPLoc paper. The paper also describes how to create a natural organically expanding anchor system that is easy to deploy and easy to maintain. The tags use a new TDoA formulation with clock drift mitigation, which is a key innovation in this paper. A big plus is that the infrastructure does not know your location. #indoor_localization, #scalable_localization

Two fixed anchors, single fast-moving tag:

Sometimes, a particular object is so important that localization of that one object is all that we desire. In a constrained cricket field setting, we show how to track a cricket ball using the wickets as anchors. We use a very fast active-passive ranging scheme to locate the ball. The physics of the ball in free fall is useful as well to derive the trajectory. The iBall paper is worth a look if you are exploring something similar. #sports_analytics, #tracking_fast_movements

One moving anchor, multiple moving tags:

A moving anchor is a bit odd. However, systems like this exist if the anchor is on a drone or a car, for example. It becomes particularly helpful if the trajectory of the moving anchor is known. We built a system for tracking firefighters inside a building using a drone flying outside. Read the TrackIO paper if you want to try something similar. #firefighters, #firstresponders, #drone

Moving Anchors and Moving Tags

An adhoc formed geometry on a sports field for example treats tags and anchors interchangeably. The P2PLoc system is a good example.  This paper uses a cascaded pipelined TWR protocol. #sports, #players, #topologies

Spoofing detection system

In a setting where the clients have incentive in cheating about their location, we have developed an unspoofable ranging protocol. Read the UnSpoof paper for insights about such a system. #trustworthy_localization, #uwb_spoofing

Inter-node distance measurement 

Sometimes, just peer to peer distance measurement is desired, and not so much location. ViSig identifies body signals such as those used by law-enforcement officials, sports umpires and referees, aircraft marshallers, construction workers, etc. Sometimes, the number of devices is very large but you care about only the few that are nearby. That needs rethinking the link-layer protocol: 6-Fit-A-Part is a system to keep a certain distance between patrons, during the Covid-19 pandemic, and is useful for grocery stores, hospitals, tourist spots, etc. #UWB_distance, #Covid-19, #body_signals

Fine-grained location

A pen's movements over a whiteboard are tracked in this fine-grained localization paper. It is called ITrackU. A small work for drone landing also used similar idea and was shown as a poster: home-coming. #education, #precise_motion_tracking

UWB as a Wireless Sensor

If you only want a sensing based system, where signal reflections or signal refractions are captured, you might want to read the LiquID paper which identifies liquids using UWB signals. UWB reflections can be used to create art! Read the Travelogue paper. This idea was exhibited at Georgia Tech's library. #uwb_sensor, #wireless_sensing

Thursday, September 23, 2021

Ultra-wideband is for everyone!

Since you have arrived here, chances are you have already heard about ultra-wideband or UWB. Most likely you heard that the latest iPhone and AirTags, or Samsung phone has UWB and you are intrigued by this technology which can be used to find your lost keys under your couch! Super useful, but didn't we hear about such capabilities before? It did not really work that well before; is there reason to expect a better product this time? The short answer is "yes!". The latest promise is brought to you by a fundamental capability that did not exist before--measuring distance accurately using wireless signals. In this first post, I would like to give a simple understanding into this new technology and why you should be fascinated by it.

What is the science behind this lost-and-found capability?

The dream goal is to be able to find any tracked object anywhere inside your home by just using your phone and asking it to "find" that object. Think about asking Siri: "Hey Siri, where are my keys?" or asking Google Assistant: "Ok Google, where did I leave my glasses?" and a map of your home pops up on the mobile screen that shows you where the keys or glasses are. The underlying technology is localization. Just like GPS that helps localize a mobile phone outdoors, indoor localization technology helps locate objects or tags in indoor spaces. Primarily they are based on a simple idea. If we can tell the distance and angle between the mobile phone and the tracked object, then we can pin-point the location of the object and guide the user to reach it. Furthermore, if the mobile phone is itself localized relative to the corners of a house, it is possible to first orient the mobile phone on the blueprint of the house, and then point to the correct location of the tracked object within the house. At its core, then, this lost-and-found functionality requires the ability to wirelessly measure distances and angles. This brings us to the next question:

How can distance be measured using wireless signals?

In theory this is simple. Suppose you have two wireless devices and you wish to find the distance between this device-pair. One device sends a wireless signal and records the time at which the signal was sent. The receiver captures this signal and records the time at which the signal was received. We then subtract the two times and multiply it by the speed of the signal to obtain the distance between the device-pair. Simple, right? You would use this method to calculate the distance between your house and a friend's house when driving over a straight path. But this idea is error-prone in the case of wireless signals because the signals travel extremely fast, at the speed of light (at a mindboggling 186,000 miles/sec or 300,000,000 meter/s). This means if we make even a slight error in recording the times, the distance will be off by a large value. So we need extremely accurate clocks (to the tune of picoseconds (10^(-12) seconds)). Furthermore, we need the two devices to agree exactly on what the current time is. Doing so is very difficult because clocks drift over time. However, instead of a single wireless signal transmission if we perform three transmissions, two in one direction and a third transmission in the other direction, it is actually possible to remove both the clock offset (initial disagreement of clocks) and the clock drift issues. This 3-message exchange is exactly what UWB does to provide a distance estimate. It is called two-way ranging since both devices are involved in the ranging (or distance measurement) process. The three messages exchanged between an initiator and a responder are typically called "Poll", "Response", and "Final". Here is a timeline of the message exchanges: 

Seems easy! What is the big deal with the ultra-wideband? Can we do the same with Wi-Fi or Bluetooth signals?

 No. We cannot achieve the same precision with Wi-Fi or Bluetooth. Recall that for wireless distance measurements to work, we need very precise time-keeping. A receiver must accurately document when exactly the signal arrived at it. To record time exactly, we need a wireless pulse that is quite sharp. The less the bandwidth of a wireless signal, the less sharp the pulse is. In contrast, the more the bandwidth of a wireless signal, it starts to approach what looks like a square wave more and more. Thus the signal rises sharply from the ambient noise and it arrival time can be exactly determined with a fast enough sampler and clock. Take a look at the following noisy signal. See how difficult it is to determine the exact arrival time of this signal:

Wi-Fi signals have a narrower bandwidth (from 20MHz to 160MHz depending on the exact Wi-Fi variant). In comparison, UWB uses signals with 500MHz or higher bandwidths. This means the signal pulse much more resembles a square wave and therefore makes precise time measurements feasible. This is why companies desirous of localization are seeking UWB technology. 

Sounds great. But you said UWB is for everyone. I don't know how I will put it to any use in my life. Can you give examples?

Maybe you have your life super organized and you do not lose things around the house all the time. But do not brush aside UWB just yet. There are things UWB could do for you that are beyond your wildest imagination. For example, a UWB mobile phone could automatically unlock your car as you walk to it. Ditto with your home door. These applications are already being looked at by automobile manufacturers and phone companies. However, the utility of UWB can go even beyond every-day convenience. My group at Georgia Tech has been involved with several projects developing new technology using UWB. We have shown that balls embedded with UWB could help make sports analytics affordable even for K-12 schools. Tracking of sports-persons can also be enabled using UWB. It is possible to track a teacher's pen on a whiteboard and digitize all the writing without requiring any special touch-screen whiteboard or a special stylus. It is also possible to identify liquids by carefully analyzing the UWB signals passing through the liquids. We have developed technology to warn if two or more shopping carts are within 6 feet of each other to help maintain social distancing to mitigate the spread of COVID-19. The same technology can also enable new types of fun sports for K-12 kids. The ability to perform indoor localization is crucial for first responder safety and we have shown that drones equipped with UWB can help locate and rescue firefighters trapped in unsafe buildings. The same technology could also enable drone-based delivery of goods by solving problems around privacy preserving landing of drones. New applications are being developed almost on a continuous basis, both at my lab as well as elsewhere. While we focus on scientific innovations in academia, industry is also moving ahead with commercializing some very interesting localization applications. The possibilities are endless. 

Various Anchor-Tag Configurations in UWB

 In UWB literature, there is a notion of tag devices being mobile, lightweight, energy constrained UWB devices that typically users or robo...