An RFID-based AT Switch – Proof of Concept

Introduction

Assistive technology switches generally fall into two categories: wired switches and Bluetooth-based wireless switches:

A Wired Switch ($60 – $90)	A Bluetooth Wireless Switch ($150 – $200)

Wired AT switches are relatively low-cost and low maintenance but require the routing of a wire of fixed length.  It’s difficult to mount a wired switch on an individual or on their wheelchair without immediately limiting their mobility. In addition, in order to use a wired switch to interact with a computer, one normally needs to include a switch adapter that maps the switch-press to a keyboard key-press. 

Almost all wireless AT switches utilize the Bluetooth communication protocol.  A few use infrared (IR) or radio frequency (RF) technologies. Bluetooth, IR, and RF transmission components are powerful transmission protocols and they carry significant costs to add them to a switch. In reality, they are too powerful for this purpose. At 24 Mbps, the data bandwidth for Bluetooth is theoretically wide enough to carry a video signal.  Traditional wireless switches are much more expensive, are much more difficult to configure, and are much more difficult to maintain. An AT switch has a very narrow signal. The circuit is just “open” or ”closed” and that one bit of information changes at a very slow rate – if at all. There is also the cost and maintenance burden of purchasing and replacing the batteries that drive these power-hungry technologies.

The question then comes to mind: “Is it possible to provide wireless transmission with low- or no-power requirements?” One possibility might be RFID technology.

RFID technology is ubiquitous. It is commonly used to automatically monitor the progress of items through manufacturing or delivery processes. The technology is comprised of two components. An RFID tag that is attached to the item of interest and an RFID reader that is constantly sending out a signal looking for any tag. The tag is composed of a chip connected to an antenna:

When a tag comes within range of the reader’s signal, the signal is picked up by the antenna in the tag which sends power to the chip. The chip then broadcasts its ID using the power of the reader’s signal. As a result, the RFID tag contains and needs no internal power. RFID readers are proximity sensors meaning that the tag has to come within range of the sensor.  [The distance between the tag and the sensor is determined by the specific RFID technology used.]

The tag reader then communicates the presence of the tag to a computer system that can use that information:

So how would a proximity technology be modified to represent circuit closure in a switch? The answer could be in the relationship between the chip and its antenna in the RFID tag. In order to generate a return signal, the RFID chip must receive a signal and therefore power from the reader via the tag’s antenna.  If a tag was initially electromagnetically shielded until the switch was pressed, and the shielding was then removed, the tag would suddenly be visible to the reader.  [Note that it is almost as effective to take the opposite action – to leave the tag normally unshielded (or intact) and only shield it when the switch is activated as in the picture below.] Another way would be to electrically isolate the chip from the antenna until the switch is pressed. 

In this design, pressing the activation surface causes the RFID tag to be shielded.

Note that an RFID tag’s ID is up to 96 bits in length, meaning that it could have approximately 2⁹⁶ = 80,000,000,000,000,000,000,000,000,000 (i.e., 80 octillion) values. That number’s almost twice as large as the number of grains of sand on the surface of the earth – squared – meaning that a room could be filled to the ceiling with RFID switches and the reader could easily distinguish the activation of one switch from activations of all the others. Another item of note is that RFID tags are routinely printed at pennies per tag making this an extremely cost effective technology.

Each time an RFID switch was activated, the RFID tag would suddenly appear as present to the RFID reader, which would pass that information on to a processor that mapped the tag’s ID to a function on the computer. When the switch was released, the RFID tag would disappear and the reader would pass that information on as well. The only component in this system that would need to be powered would be the RFID reader which could be mounted somewhere near the individual with a disability – for example, on their wheelchair, and draw power from the wheelchair battery or mounted on a desk and draw power from a power strip next to a computer.

Types of RFID Technology

We’ve all experienced RFID technology in action.  If you’ve had your dog “chipped”, the vet inserted a small RFID tag under the dog’s skin in the area of its neck.  If your dog is ever lost and taken to a vet or shelter, the personnel there can pass a wand over the dog’s back and read the ID value of the take under the dog’s skin.  The wand has a built-in RFID reader and it will display the ID of the tag.  That value can then be compared to a database of values to look up your name and phone number.

If you’ve ever heard an alarm go off when someone leaves a store, it’s likely that the individual passed an RFID reader at the store’s exit holding an item that they had not “purchased” – meaning that the RFID tag embedded in that item had not had its ID recorded as “purchased” by the store computer.

The primary difference in RFID technologies between these two stories is the distance over which the tag can be read.  In the first example, the tag has to come within a few centimeters of the reader.  In the second story, the reader may be able to recognize the presence of the tag a meter or more away from the antenna.  The greater the distance between the tag and the antenna, the more powerful (and expensive) the reader will be.  More about the different RFID technologies, here.

RFID tags can be passive or active.  A passive tag has only an antenna and chip.  An active tag has an embedded battery, as well, and can be recognized at a longer distance from the reader.  As you would expect, an active tag is more expensive than a passive tag.  More about passive and active tags, here.

The optimal RFID system for an RFID-based AT switch, would involve an inexpensive reader that can read passive tags at a distance of 3 or 4 meters.

Those kinds of distances require a reader that transmits in the ultra high frequency (UHF) range.  Passive tag UHF readers have the following characteristics:

• Primary Frequency Ranges: 860 – 960 MHz
• Read Range: Near Contact – 25 Meters
• Average Cost Per Tag: $0.09 – $20.00
• Applications: Supply Chain Tracking, Manufacturing, Pharmaceuticals, Electronic Tolling, Inventory Tracking, Race Timing, Asset Tracking
• Pros: Long Read Range, Low Cost Per Tag, Wide Variety of Tag Sizes and Shapes, Global Standards, High Data Transmission Rates
• Cons: High Equipment Costs, Moderate Memory Capacity, High Interference from Metal and Liquids

A goal in identifying candidate hardware will be to avoid the Con of “High Equipment Costs”.  Otherwise, the read range and tag costs appear very promising.

Developing an RFID-based AT Switch

Several steps will be necessary to prove-in the concept of an RFID-based AT switch:

1. Document requirements.
2. Select candidate hardware.
3. Experiment with simple shielding techniques.
4. Design a switch that can shield/unshield a tag when the activation surface is pressed.
5. Develop a software architecture that can take an RFID tag ID appearance and turn it into a control action on a computer.
6. Extend the software architecture to support multiple switches and multiple tags.
7. Extend the software and hardware architecture to support multiple switches, multiple tags, and multiple computers.
8. Select additional hardware to support multi-computer communication.
9. Develop the individual software components.
10. Develop documentation and training for the final system
11. Explore designs that expose and hide the tag via non-shielding techniques.

Requirements

There are three core requirements associated with this solution as with every assistive technology device designed and developed by Volksswitch.org:

1. Every design (non-commercial hardware and software) shall be placed in the public domain.  No one shall be charged for access to the design and its documentation or training.  The user is responsible for purchasing any commercial components.
2. Customization and personalization shall be baked into the design.  Individuals with disabilities have very specific needs that cannot be met by a one-size-fits-all design mindset.
3. Procurement of components and assembly shall be well within the abilities and means of individuals with disabilities, their caregivers, and therapists.  [There shall be no need to involve a maker/geek/nerd.  The role of a maker/geek/nerd is to create the design, documentation and training – not to sit in the critical path to success.]  In other words:

• • The design shall involve off-the shelf components.  [One exception is that some components may be 3D printed as an option.  3D printable components shall be designed such that they can be simply sliced and printed using PLA.  If absolutely necessary, TPU may be specified.  Avoid the need for supports.]
  • Assembly of the hardware and software components shall be easy and supported by clear and complete documentation and training.  [“Easy” is defined as requiring no special skills beyond gluing, bolting, snapping, and running an installation program. Where 3D-printed parts are involved, minimal slicing and printing skills are involved.]
  • Troubleshooting the assembly and operation shall be simple and require no special skills.
  • Corollary: Unless the design is implementable at a significant cost savings, or offers significant new features, over commercially available designs, it should be abandoned in favor of the commercial product.

Additional Requirements for the RFID-based AT SwitcH

1. The cost of the system (including 2 switches) shall not exceed $300.
2. The system shall initially support 2 switches and one computer.
3. The system shall eventually support at least 50 simultaneous switches and 10 simultaneous computers.
4. The system may initially support only Windows-based computers but the design should not preclude adding support for Mac OS, iOS, and IoT systems in the future.
5. The system shall support communication distances of at least two meters between the switch and the computer that is being controlled.
6. The system shall initially support a pressure-type AT switch.  Exploration of other switch types will take place in the future and shall not be precluded by the system design.
7. Each switch shall support several different mounting options.
8. The cost of constructing a single switch shall not exceed $4 (plastic and tag).
9. Once a switch has been assembled and is functioning it shall require no further maintenance.
10. The system shall provide a simple user interface for mapping (and remapping) a tag ID to an action on the target computers.
11. The system shall support simple troubleshooting and maintenance through replacement of parts.

Candidate Hardware

After some research, we located two promising candidates for development of the RFID reader.  The first is from SparkFun and the second is from ThingMagic:

SparkFun	ThingMagic
SparkFun Simultaneous RFID Reader – M6E Nano	ThingMagic RAIN Starter Kit
Cost: $224.95	Cost: $525
Includes:  The board and a piece of gray, non-conductive Thermal Gap Filler to help with both heat dissipation and to cover the exposed ground plane (to prevent circuits from shorting against it)	Includes: USB Pro RFID Reader unit USB Micro-B connector, with removable 6 ft cable with dual USB-A type plug One year factory warranty
Features: EPCglobal Gen 2 (ISO 18000-6C) with nominal backscatter rate of 250kbps Separate read and write levels, command adjustable from 0dBm to 27dBm in 0.01 dB steps 0.84W in ready mode 0.00025W in shutdown mode Up to 150 tags/sec to read 96-bit EPC 80msec typical for standard write of 96-bit EPC	Features: Support for EPC Gen2V2 (ISO18000-63), AEI ATA, ISO 18000-6B, IPx protocols. Powered by PC, laptop or tablet through a host USB interface. Windows, Android or Mac host operating system. Support in the 865-869 and 902-928 UHF RFID carrier frequency ranges to accommodate worldwide regulations. Certifiable for operation in countries adopting FCC, ETSI, India, Korea, Australia, P.R. China, and Japan regulatory specifications (Open (Customizable) 865-868 MHz and 902-928 MHz). Internal linear polarized antenna with peak gain of 1 dBi. Additional antenna port for 2nd external antenna. ThingMagic Reader Assistant software and applications development tools. 50 tags/second read rate. High impact plastic case.
Other components needed and their cost: UHF RFID Antenna TNC(?): $38 RP-SMA to U.FL Interface Cable(?): $5 Interface Cable for TNC to RP-SMA(?): $6 Serial to USB Bridge: $8 Break-away Headers: $2 UHF RFID Sample Tags: $2 UHF RFID Sample Tags: $2	Other components needed and their cost: circular polarized antenna: $64? interface cable: $?
Micro-controller or Micro-processor support: Arduino (requires additional headers be purchased and installed)	Micro-controller or Micro-processor support:
Documentation: Schematic Eagle File RFID basics Hookup Guide Product Brief Datasheet Design Guide Thermal Gap Filler Firmware Update Firmware release notes Product Page OEM Documentation page Universal Reader Assistant (Windows only) GitHub (Design Files) GitHub (Arduino Library)	Documentation: ThingMagic® USB Pro UHF Reader

INITIAL HARDWARE TEST RESULTS

 

Experimenting with Simple Shielding Techniques

This step will be critical to determining a feasible switch design. 

1. What materials can be used to shield the tag?  Preference should be given to materials (e.g., aluminum foil) which are readily available to everyone.
2. How much or how little of the tag needs to be shielded in order to hide it from the reader?  Passive UHF tags that can be read at long distances tend to have large antennae. If it’s only necessary to shield a small part of the antenna to make the tag invisible, there will be many more options for the mechanical design of the switch.
3. Could something as simple as tag orientation or bending affect the tag’s readability? In some ways this could make the design more difficult since it’s hard to predict the orientation with which the switch will be mounted.

Initial Shielding Test Results

 

Switch Design

Given options for ways to make the tag appear and disappear to the reader, we need to come up with a mechanical design for the switch.  If possible, the switch design should meet the goals of the Volksswitch Proof of Concept:

• modular design with an independent activation surface
• customizable activation pressures (using a flexure spring)
• customizable activation distances
• customizable mounting options
• snap-together components

The Round Flexure Switch – 60 mm from Makers Making Change incorporates several of these same goals.

Initial Switch Design

 

Single-switch, Single-computer Software Architecture

This is the “proof of concept” for the system.  Once the hardware is tested using the supplier-provided test code, a software architecture will be needed which identifies the tasks needed to map a recognized tag ID to a computer keyboard action and where in the hardware component chain that the accompanying software should reside.  Even at this point, simple software installation support should be considered.

Multi-switch, Multi-computer Software Architecture

The architecture and associated software will first need to be extended to support multiple switches transmitting unique tag IDs.  In order to support multiple computers, there must be a way to connect the reader and its supporting micro-processor to the local network.  This step should take place in tandem with the design of a network architecture.  One key question will be: should switch activity be centrally routed to a specific target computer or should each computer listen to the activity of all switches and respond to any switch that it wants (potentially with different actions on each computer).

Network Architecture

There will need to be some experimentation with network communications options.  WiFi routing seems appropriate for this application but Ethernet may provide the better routing support.  Or maybe this is where Bluetooth comes back in – not at the switch level but at the reader level.

Hardware Candidates for Networking the RFID Reader

Some early decisions, like adding a Raspberry Pi to the reader may facilitate WiFi, Ethernet, or Bluetooth routing.

Develop Components of the Software Architecture

This task is actually taking place at several steps in the process.  As the hardware and software architectures firm up, individual software module boundaries, communication interfaces, and programming language decisions will emerge.  A testing strategy will be also be needed.  As mentioned earlier, a plan for simple, end-user software installation and troubleshooting should be considered early in software design.

Documentation and Training

The procurement, assembly, troubleshooting, and operation of the system must be clearly and completely described at a level sufficiently simple for a disabled individual, their caregiver, or a therapist to perform.  Accessing the documentation and training must be as simple as going to a web page or a YouTube video.

Non-Shielding-based Techniques

One of the limiting factors of long range, passive RFID tags is that they have large antennae. Since only the chip changes from tag to tag, it would theoretically be possible to reuse the same antenna with multiple chips – say, in a switch with a single housing and multiple activation surfaces – by electrically connecting a particular chip a shared antenna depending on which activation surface was pressed.  Another application would be to embed a microswitch in the chip to antenna interface and then use a standard switch design to translate pressing on the activation surface to engaging the microswitch allowing for more switch designs:

Since the length of the antenna likely plays an important role in determining the associated radio frequency these approaches may be difficult to achieve.