Slightly off topic, for fun: what happens when we move from 15 USB interconnects, to say, 100 million? What happen when you design your own (array of) USB host controller that are/is not limited to standards-based 5-layer fanout: assume your packet protocol on the serial bus can be tweaked.
Lets now imagine a football field size arena of ICCs mounted on reel/tapes, streaming through readers, powered on the reel/line with RF, with internal clocks. Think big, like a giant collector array, where streaming address the heat problem and collects the offline-generated results. Like the Hoover Dam, Panama Canal, Glomar Explorer: breaking raw ciphertext and potential keys has the same value, motivating the same infrastructure (and secret budgets). (You would need an intelligence agency with its own onsite fab to hide this silicon output, economically.)
Putting the USB SIE into the ICC, i.e. eliminating the "reader", has had no effect on the price of the interconnect with PC, in the banking market: its still $30 retail, volume! But setup time goes down.... Javacards (or MEL) would be a boon for ICC-array reconfiguration: downloading applets with new native code to drive the cryptocoprocessor, as algorithm targets change, by intelligence prio!
rity. lets call this "Array Reconfiguration"
In 1996, rumour has it, a 3des-grade key/ciphertext had a 5 hour lifespan... Must be down to 5s, by now (derived rumour). We need to think big, really big, when designing ICC interconnects. When considering counter-counterintelligence strategies, we also have to think about exploiting means to cause array reprogramming, to force errors, confuse the operators, deceive the intelligence officers makin the targeting calls, break the engineering assumptions of the bus, and..ultimately... widen crypto windows.
Get a FREE online virus check for your PC here, from McAfee. _______________________________________________ Muscle mailing list [EMAIL PROTECTED] http://lists.musclecard.com/mailman/listinfo/muscle
