300 feet, or even 10 feet, of outside penetration would be very unacceptable. Would you like to be tagged with an extra $15 nuisance fee on your cell bill just because you walked (or even drove) past a restaurant while using your phone?

As for cost, I think the triangulating system would be far less expensive, assuming it can be made to work. As you said, the cost of the equipment would be only a couple or few hundreds of dollars. Moreover, if the restaurants are going to start collecting on those calls they will trample each other to buy the system. But $0.31 per unit multiplied 10,000,000 times is millions of dollars. Then there is the cost of legislation to require it. And the cost of confiscating all of the older phones without the circuit, unless you just want to let them not be subject to the nuisance fees. That would certainly increease the market value of older technology phones.

By using a system that triangulates the position of the caller (inside or outside of the protected area), the need for, and expense of, specially equipped phones is avoided altogether. The question then becomes "How small a time increment is required to implement this type of system?"

In order to measure a difference of 1 m one would need to measure a time increment of (1 m) / (3E8 m/s) = 3 ns. Assuming one period per increment of the timer this would require a 300 MHz clock. To measure a difference of 1 cm a time increment of (1E-2)/(3E8) = 33.3 ps would be required (similarly implying a 30 GHz clock).

This suggests to me that this couldn't be done directly with any '51 derivatives that I know of. However, I suppose one could employ special ripple counters for the actual measurements and then just use a '51 to run the show.