I am working on a very non-conventional proposal. One of my probable clients wants us to develop an entire digital Scrambler Module for Motorola Professional Series Radio. He already has an analog version of scrambler and he wants a digital one. Initially I had the information that he needs a Cryptographer who could design a digital scrambling algorithm for him. We have a cryptography oriented team that we had cultivated since inception, so I thought this would be good work for my cryptography team, and I approached the probable client. After discussions, we realised that he needed not only the algorithm, but the entire associated system that would lead him to put the algorithm into a chip that could be mounted on the radio. So we offered the client development of algorithm and consulting services for burning the algorithm into a mountable chip. The proposal had three stages: Development of algorithm that could meet specific performance requirements of a handheld class application with adequate security, consulting services for testing and prototyping the algorithm into a chip. Once the prototyping is done, we offered consulting services for selection of a vendor who could mass manufacture the chip. We quoted a price that factored in charges of our consulting services. Actual costs involving hardware components would have to be borne by the client.

The client came back to us and said he wanted an all inclusive price consisting of consulting costs, hardware component costs, software development costs, chip fabrication costs, tool costs and manufacturing costs. He said that he would just pay money and every other headache would be ours. But he would want to know the cost upfront. Our core competency in this case was just cryptogrphic system design, and we had offered manufacturing related consulting services as a value add. We don't know the actual costs and we don't know how many iterations of manufacturing cycle will go on before we perfect the product. Its a very risky engagement. Before puttiing up a revised proposal we need to know every single step and every single cost. We have to factor in unforceable events, and price of it. At the end of it the number should not scare the probable customer. Otherwise he will postpone the decision.

We decided to go ahead with the proposal, and now we are researching what it takes to meet the requirements. Basically the work we should have done after the engagement award is being done before proposal. But we want to do it. This offers us a chance to learn what is takes to bring a security design into hardware. We have another product called ControlCase (www.controlcase.com), and this engagement could help us getting a hardware version of it. We will learn the entire VLSI cycle and possible pitfalls.

We started with getting papers on existing security algorithms being put on chip, and doing research on FPGA (Field Programmable Gate Array)/CMOS based VLSI implementations. We have the rough outline of emerging algorithm, and we now know the coding language for electronic design(VHDL and Verilog). We understand the VHDL->RTL->Simulation->Synthesis->PlugNRoute cycle for FPGAs. I have even tried my hand on VHDL coding to get a feel. So we have outline of core areas.

But going through other details, I am getting scares. We don't have full listing of analog scrambling functions in old scrambler. Its just not security but functions like ANI at the end of TX, No-mute during lead-in delay, silent signalling, data operated squelch, and a couple of others. We don't even have the signal diagram of old scrambler. Its just not an algorithm, but a full system that has to be put on the digital scrambling board whose specifications are not known to us. If we don't even know what we are buiding, how can we build it? The client does not know about it, and Motorola will never tell us its internal signalling/function formats. But yet, I want to do it. I took an evening off yesterday so that I could get my mind off..but today the complexity of problems remain, and no solution in sight.