If you have a power supply on the measuring side to power an LED, surely you have a power supply on the measuring side to feed a variety of small active circuits, which control that LED based on a much smaller current through the thermistor. Starting with the comparator-based schematic in
https://www.eevblog.com/forum/projects/pwm-signal-from-555-timer/msg17855/#msg17855 you would replace R5 (or R7) with the thermistor and R7 (or R5) with a fixed resistor of 1-3x the thermistor's nominal value, balancing the range of the output against self-heating. R6 can be substituted with a short circuit for the first iteration, and you can replace it with two fixed resistors to reduce the change in duty cycle per °C in case you need to increase measurement range.
On the receiving side, first figure the minimum and maximum voltages of the pseudo-triangle wave at the - input of U1B. You can treat the oscillator as two separate voltage dividers, one with Rtop=(1k+R3)||R1 and Rbottom=R2, the other with Rtop=R3 and Rbottom=(R1||R2). Next, measure the duty cycle of the received signal by your favorite method, then map 0% duty to the minimum voltage and 100% duty to the maximum voltage. Now you know the voltage on the + input as seen by U1B, and you can now solve the R5-R6-R7 voltage divider for Rtherm, from which you can look up or calculate Ttherm by the usual equations/tables.
Between the two circuits, you can place optocouplers, discrete LEDs/phototransistors, ultrasonic transmitter/receivers, or carrier pigeons, as available.
(Tip: all the important voltages in Zero999's comparator circuit are proportional to Vcc, so Vcc drops out of all the equations and its exact value is irrelevant. For the purpose of processing the received signal, you need only concern yourself with the ratio of any given voltage to Vcc. However, Vcc must not change quickly during operation, as any variations over short-time will be coupled into the transmitter output.)