Mechanism for Unique Decipherability of Morse coded message for low cost ubiquitous bi-directional communication with Deaf-Blind & its extension to ubiquitous mobile phones.
Deaf-blindness is a rare condition in which an individual has combined hearing and vision loss, thus limiting access to both auditory and visual information.
Education: Persons with deaf-blindness may be accompanied by an intervenor, a professional who is trained in tactile sign language. This sign language involves touching the hands of the client using a two-handed, manual alphabet, also known as finger spelling. This is very important at it is through this are they exposed to concept of words and their association with things present in surroundings.
Methods of communication include:
- Tactile signing, sign language, or a manual alphabet such as the American Manual Alphabet or Deaf-blind Alphabet (also known as “two-hand manual”) with tactile or visual modifications.
- Interpreting services (such as sign language interpreters or communication aides).
- Communication devices such as Tellatouch or its computerized versions known as the TeleBraille and Screen Braille Communicator which are very costly >1500 $.
In general Communication via any of the above methods is limited to select few who know the tactile language that person has been trained in, creating a severe barrier for free exchange of feelings and expressions with others. There have been attempts to solve this problem to an extent with Morse Code.
Limitations in present Implementations of using Morse Code
Morse Code: The Morse code fits very nicely as solution to above communication barrier as it can be interpreted via sense of touch. Now we will discuss issues that pose challenge to its application and then will discuss solutions to them.
Issue1- Absence of any physical identification of input area The Bapsi (Bi-Directional Access Promotion of Society of India) had created amazing vibration series of products in 2012 that uses Morse code to encode message in form of vibrations of phone to convey message to deafblind and used graffiti keyboard to write text. Actually there have been multiple works on using Morse code for deafblind communication since time of Helen Keller. But current implementations are frought with serious limitations that limits its use.
We all have experienced issues with touchscreen like wrong input, finger not detected due to callouses and other issues. Imagine how difficult it is for deafblind to use graffiti keyboard as it is impossible for them to see if they are crossing over to number/text area & if the orientation is correct.
Issue2 - In using Morse coded input, Morse code cannot be used straight forwardly for input because it is not uniquely decodable & thus results in ambiguous messages.
Issue3 - Tactile Feedback: Any device that is to support deaf-blind communication should provide tactile feedback and should not cause any errors due to unintentional strokes.
Sol to Unique Decipherability Problem:
Basic Definitions:
·A code C is just a finite set of finite length strings, called codewords. For example, C = {aba, a, abb, ab, ba} is a code, and aba is one of the codewords in C.
· A message M is a string created by concatenating strings from C, with repetitions of codewords allowed. For example the string “abaabbababbaa” is a message that can be created using the codewords in C. A code C is called Uniquely Decipherable if and only if every message that can be created from C is UD.
Approach: Given C, we build a graph G with one node for each string that occurs as a suffix of any codeword in C, including a node for each complete codeword (the trivial suffix). Note that the same string might occur as a suffix of more than one codeword, but it is only represented by a single node in G. We label each node by the suffix (string) it represents.
In graph G, there is a directed edge from node u to node v
- if there is a codeword c such that c = uv (this is an L1 or Type 1 edge), or
- if u = cv (this is an L2 or Type 2 edge).
Figure 1 shows graph G for the code C = {ac, bc, bca, cb, bacb}.
Theorem: C is not UD if and only if there is a directed path in G with at least one edge, starting at a node representing a codeword, and ending at a node representing a codeword. The start and end nodes can be the same, but don’t have to be.
Proof I will not discuss here as it is a bit involved , I am indebted to prof Dan Gusfield for his lecture on Unique Decipherability.
Unique Decipherability Issue in Morse code
Developed by Samuel Morse, In Morse code each letter is translated into sequences of short pulses(dots) and long pulses(dashes) lets represent them by 0 & 1 for our convenience.
More frequent letters are assigned shorter sequences for efficient communication. Ex e =>0, t=>1, a=> 01 . Morse code is a variable length encoding, due to use of such short sequences the encoding of words becomes ambiguous.
Ambiguity : From our Example C = {e = 0, t =1, a=01} subset of morse code above, let Message M = 0101 can be interpreted as any sequences of letters
M1 i.e eta = 0,1,01 or
M2 i.e aa = 01,01 or
M3 i.e etet =0,1,0,1 or
M4 i.e aet = 01,0,1 are all valid decoding of received string message =0101
Actually Morse code transmissions involve timing as an integral part of transmission mechanism to solve ambiguity during which transmitter would remain idle.
This dependency on time makes it tough for deafblind & is main challenge for even normally sighted people learning morse. to keep track of leading to ill formed message. Even Google’s implementation of morse input has this limitation as shown below.
Solution:
- If we can add a mechanism for delimiting the codeword, we will be done. So instead of a binary coded Morse I propose a ternary coded Morse that has {0, _, 1} as its symbols for resolving ambiguity by using an explicit transition signal marked by _ , which is only present at end of each codeword. This will resolve issue no 1.
2. To provide a tactile interface that can give a feedback as well as intuitive input area for Deafblind, I propose to extend a 3 physical momentary button interface this will solve issues 1&3 highlighted above. But this adds a new issue 4 how do we add physical buttons with tactile feedback ? As nowadays phones have capacitive physical buttons or touchscreen only input.
3. Solution to issue 4 We can exploit the legacy 3.5mm Audio Jack and use the android accessory specification that can be overridden using external circuitry to add 3 physical buttons needed for our transformed Morse. This will solve issue of tactile feedback and user will always know what he typed and what is registered as input to system.
If above changes into input mechanism are made any mobile can be used as ubiquitous bi-directional communication system to communicate with deaf — blind.
Now we can capture this media button events using a broadcast listener to capture the above media button events and pass them to parser to convert it to equivalent symbols.
I am indebted to project advisor Prof Satya Prakash Sahu who mentored me during my major project in HCI as well as agreeing guide on this project also.
