Purpose of the Gaussian filter in GMSK
It allows the signal controlling the frequency to take smooth intermediate values, rather than jumping directly between +1 and -1.
For example, without filtering:
Bit: 1 0
NRZ: +1 ─────┐
└──── -1
With Gaussian filtering, you might get something like:
Filtered: 1.0 → 0.95 → 0.8 → 0.4 → 0.0 → -0.4 → -0.78 → -0.95 → -1.0
Why do we want those intermediate values?
Because those values control the instantaneous frequency.
Your MSK code essentially does:
frequency = carrierFrequency + frequencyDeviation * bitValue;
For MSK:
bitValue = +1
↓
frequency = fc + Rb/4
or:
bitValue = -1
↓
frequency = fc - Rb/4
There are only two frequencies.
With GMSK, suppose:
filteredBit = 0.8
Then:
If:
filteredBit = 0.2
then:
And if:
filteredBit = -0.78
then:
GMSK = Gaussian Minimum Shift Keying
MSK → minimum frequency separation / continuous phase
G → Gaussian filtering is applied before the MSK frequency modulation
The Gaussian filter essentially spreads out/smooths the bit transitions, which reduces rapid changes in the signal and therefore reduces spectral sidelobes / occupied bandwidth.
One important point: the filter doesn't mean the transmitted bits themselves become 0.8, 0.9, etc. The original bits remain 0 and 1. The intermediate values exist in the filtered waveform that controls the instantaneous frequency/phase.
Summary
The Gaussian filter creates intermediate values between the NRZ levels, and those intermediate values make the frequency transition smooth.