Voice mail technology has its demerits the biggest of which is lack of privacy. As more and more people continue to use the system, the threat continues to grow daily. This paper suggests a voice privacy system which would guarantee privacy. The system will not only optimize the bandwidth but also enhance privacy in sensitive organizations like banks
Statement of the problem
Each time you use the telephone or the voice mail, you are risking being overhead by an eavesdropper. While the implications may not be too high for the home use, the case is different for the business environment like the bank. We have already heard of sensitive information being lost due to eavesdropping on phone calls between business magnates that have resulted in huge losses or scandals.
While using of the telephone to make business transactions results into security loopholes, the proposed model has a solution to this. It makes it possible to have privacy guaranteed. The model is built on the foundation of wavelets. The model is able to transform sound at the sending node and the receiving node will recover the sound by making an inverse transformation. It uses the mechanism of Discrete Wavelet Transformation (DWT) to achieve this (Poliker).
Traditionally, privacy of systems was considered to be a military prerequisite. The concept of privacy in the commercial systems developed much later. Voice mail is a product that is provided by many vendors and adding a form of privacy to it would make it even more palatable. The telecommunication industry is suffering great losses due issues related to eavesdropping.
Wavelet analysis is done by breaking down of the signal into segments (wavelets). The wavelet has very irregular shapes and they are also compactly supported. This is desirable qualities in analyzing of signals that are not stationary. The irregular nature of the wavelets makes it possible for them to analyze a signal that is discontinuous and one which has acute changes. The feature of compactly supported allows the wavelet makes it possible to localize signal features.
The main idea in the proposed model is doing away with the irregular transformation. Instead the model will use discrete wavelet transformation (DWT). This will result in five dissimilar bandwidths of frequencies 0-250, 250-500, 500-1000, 1000-2000, 2000-40000. Since the normal telephone sound rarely exceed 2000, the 2000-4000 band is discarded. We will take an arbitrary block sample of 32 (which can be increased to achieve higher security). The speech is synthesized and translated by 250mz. On the receiving node, the speech converted back to the original band (0-2000). The sound is then read by the MATLAB application. MATLAB is universally used for processing of digital data (Stormy, 147). The figure below summarizes the process both on the sending and the receiving nodes.
January 10 January 15 February 2
initial prepation installation of system testing of system Job completion
The suggested budget is as follows:-
Purchase of equipment(cabling, Convertor, tonegenerators) 500,000
Outsourcing costs 10,000
labor costs 6,700
The proposed model will go along way in securing telephone transaction for the bank and will therefore seal the loophole that exists which could otherwise be used as a channel for eavesdropping and lead to fraudulent activity. Once implemented, the model can be adopted by other organizations including the military to secure communications across the telephone lines.