Yes, time-of-flight (TOF) or time difference of arrival (TDOA) with RF is difficult to achieve accurately. With ultrasound it is much easier because of the lower speeds involved and in particular for TDOA the ability to communicate between two nodes using a higher speed medium like RF or electrical signals on wires to achieve an accurate time reference.
As was just mentioned, having a shared time reference is important if you want to implement a reverse-GPS-like architecture (single transmitter, multiple receivers who must coordinate a time reference, using TDOA to compute position).
However, at long ranges (greater than 100 meters) you might be able to get decent results using time-of-flight by separately measuring the round-trip time (RTT) between your unknown point (the RC craft) and each reference point (one of the multiple transceivers serving as position references), then can use the individual RTT times to derive a position. You won't get high precision without specialized hardware or a custom RF ASIC, probably, and perhaps less precise than GPS but it would be a self-contained system.
Let's see: RF signals travel through air at approximately the speed of light, so assume the speed is
v=3.0 * 10**8 m/s. Suppose you have a microcontroller with a 72 MHz instruction cycle. This is about 14 ns per instruction. Then the RF signal travels 14 ns *
v = 4.2 metres per instruction cycle. Not bad! Perhaps you could achieve 10-20 metre accuracy in the best case if you have a timer/counter peripheral that can capture the exact timer count when the radio indicates the incoming packet has been received, in the optimal case. However, you rarely have a direct, full 72 MHz data bus with the radio in a normal microcontroller-based RF system, in my experience. In particular, any jitter in the RF will reduce the accuracy greatly at these levels of precision.
The
Jennic JN5148 has an interesting hardware support for time-of-flight range measurement. I have read articles about implementations of time-of-flight ranging using regular 2.4 GHz radios interfaced with a microcontroller, which I find interesting because I think it's great to be able to use general-purpose components rather than require a specific MCU, radio, etc., for each new system.