I think you need take official registered a USA or EU patent to you DB. This will be protect your Idea from stealing and maybe give some money. Then you have a patent, can try sell your method to google, microsft etc.
I don't believe a brain blood cloth is willing to wait some exponential time for a "binary" DB to be "scanned". The current state of practical technology uses bytes, words, 64-bit addressing and registers, because these are the actual smallest units of data that are being read or operated within a single clock cycle, by a CPU, controller, or whatever. Especially memory, the higher the bus width, the more data can be transferred per cycle.
Using 0 and 1s is already the basis of everything we use every day, and lots of things in CS make heavy use of them for memory-trips efficiency already, flags, algorithms, etc. but they DO NOT do this because it uses less data, they do it because IT IS FASTER. As a few already said before, "scanning" speed of said DB is sub-optimal to say the least, but hey, I'm no one to judge people's ideas. However, I must say it does not sound like something that any non-graduate CS student hasn't already "seen this, done that" as there are dozens of algorithms that do a much better job. This is why I asked about complexities or insert/update/query operations on said DB, otherwise it's not a DB but mangled packed bits, e.g. some compression output.
