Generally, the higher the value of the Signal to Noise ratio (S/N or SNR) the better. A higher S/N ratio means less likely that bits be in error. However, for every communication system a minimum value of SNR should be preserved, e.g. by increasing the power output from the transmitter or otherwise we will have a non reliable communication. The value of this minimum threshold depends on the bit error rate (BER) required for the system at the physical layer. So for every system, a certain BER is required which in turn can't be guaranteed below a certain minimum SNR. The value of this minimum threshold depends on the error correcting capability of the system (the complexity of the FEC) and the modulation type and order as follows:

1. Modulation: a binary phase shift keying (BPSK) modulation can tolerate higher SNR values than QPSK. QPSK , in turn, can tolerate higher SNR values than 16 QAM,...and son. This is because the higher the modulation order, the closer the points representing each level in the constellation space of the signals (BPSK has 2 points widely separated, QPSK has 4, ..., 256QAM has 256 points that are hardly distinguishable in the presence of noise).

2. Coding: the FEC coder can have different rates, e.g. 1/2 rate coder, 3/4 coder,...etc. The higher the coding rate of the coder the less is its capability to correct bit errors. which means that we need to use lower rates if we want a more robust system against noise to tolerate higher SNR values.

The required BER value depends on the higher layer requirements and how this BER is reflected at upper layers of the communication system. e.g. the BER increases the block error rate at the MAC layer which necessitate more requests for retransmission by the ARQ unit. This amounts to increased delays. Block error rate at MAC layer can also be reflected as increased error rate in segments, sessions,...etc at upper layers leading to different sorts of QoS impairments (low voice quality, delay, jitter,...etc). Hence, Improvements at upper layers (e.g. in new systems there is HARQ which require less re-transmissions) can result in higher rates of bits in error tolerated. Thanks. @AlDmour

The transmission power is 2 W(maximum). By considering the loopback signal from antenna of BTS, the power varied. 1W is standardised for GSM1800/GSM1900. The Frequency reuse (is limited to 35 Km) holds the S/N campansiation.