GMSK Modulation

Gaussian Minimum Shift Keying (GMSK) is a pulse-shaping extension of MSK that passes the binary data through a Gaussian low-pass filter before frequency modulation. This filtering operation smooths the rectangular data pulses, reducing abrupt phase transitions and significantly compressing the transmitted spectrum. GMSK is most famously used in the GSM (Global System for Mobile Communications) standard.

Core Concept of GMSK

In standard MSK, the frequency pulse is rectangular — the instantaneous frequency switches instantaneously between two values at bit boundaries. This abrupt switching creates high-frequency spectral components (sidelobes) in the transmitted spectrum. While MSK's sidelobes are already lower than conventional FSK, they are still large enough to cause adjacent channel interference in densely packed cellular frequency plans.

GMSK solves this by inserting a Gaussian low-pass filter between the binary data source and the FM modulator. The Gaussian filter has an impulse response that is bell-shaped (Gaussian in both time and frequency domains). It converts each rectangular bit pulse into a smooth, overlapping Gaussian pulse. The frequency modulator then produces smooth, continuous frequency transitions instead of abrupt ones.

The key design parameter in GMSK is the BT product (Bandwidth-Time product), where B is the 3-dB bandwidth of the Gaussian filter and T is the bit period (same as Tb). This single number controls the trade-off between spectral compactness and inter-symbol interference (ISI).

Mathematical Expression

The impulse response of the Gaussian filter used in GMSK is: h(t) = (sqrt(2π/ln2)) * B * exp(-2π^2 * B^2 * t^2 / ln2). This filter is applied to the NRZ bit stream before frequency modulation.

The GMSK signal retains the same modulation index h = 0.5 as MSK. The instantaneous frequency still varies between f1 and f0, but now the transitions are smooth rather than instantaneous. The phase trajectory φ(t) is the time integral of the instantaneous frequency deviation, and in GMSK this integral produces smooth curves instead of linear ramps.

The spectral efficiency of GMSK is defined by the 99% power bandwidth. For BT = 0.3, this is approximately 0.86/Tb, while for MSK (BT = infinity) it is approximately 1.18/Tb. The narrower bandwidth comes at the cost of ISI: each bit's Gaussian pulse extends into adjacent bit periods, causing inter-symbol interference.

Practical Understanding

GSM uses GMSK with BT = 0.3. This specific value was chosen after detailed system-level optimization for the 900 MHz GSM frequency plan. The 200 kHz channel spacing in GSM, combined with BT = 0.3, ensures that 99% of the transmitted power falls within the 200 kHz channel, minimizing co-channel and adjacent channel interference.

The ISI introduced by BT = 0.3 is handled at the GSM receiver using an equalizer (specifically a Viterbi equalizer). The training sequence embedded in each GSM burst allows the receiver to estimate the channel and correct for both ISI and multipath distortion. So the system is designed with the knowledge that GMSK at BT = 0.3 will introduce controlled ISI.

Like MSK, GMSK maintains a constant envelope, making it compatible with Class C nonlinear power amplifiers. This allows high power efficiency in mobile handsets, which is critical for battery life. The combination of compact spectrum and constant envelope makes GMSK one of the most practically optimized modulation schemes ever standardized.

Solved Numerical Example

For the GSM standard using GMSK with BT = 0.3 and a bit rate of 270.833 kbps, calculate the 3-dB bandwidth B of the Gaussian filter and verify the carrier frequency deviation used in GSM.

Given:
Bit Rate Rb = 270.833 kbps
Bit Period Tb = 1 / 270.833 x 10^3 = 3.692 microseconds
BT product = 0.3
Modulation Index h = 0.5

Why this formula applies:
BT = 0.3 defines the Gaussian filter bandwidth. h = 0.5 defines frequency deviation.
Frequency deviation delta_f = h / (2Tb) = 1/(4Tb)

Formula:
B = BT / Tb
delta_f = 1 / (4 x Tb)

Substitution:
B = 0.3 / (3.692 x 10^-6)
delta_f = 1 / (4 x 3.692 x 10^-6)

Calculation:
B = 81.25 kHz (3-dB bandwidth of Gaussian filter)
delta_f = 67.7 kHz

Final Answer:
Gaussian Filter 3-dB Bandwidth B = 81.25 kHz
Frequency Deviation delta_f = 67.7 kHz
Total frequency swing = 2 x delta_f = 135.4 kHz (fits within 200 kHz GSM channel)

Exam Tip: GATE questions on GMSK focus on two things — the BT product meaning and the consequence of reducing BT. A lower BT (like 0.3) gives a narrower spectrum but introduces more ISI. A higher BT approaches MSK behavior with less ISI but wider spectrum. Remember: GMSK always has h = 0.5 (same as MSK). The BT product only controls the Gaussian prefilter, not the modulation index.

GMSK vs MSK Spectrum and Phase Comparison

• GMSK inserts a Gaussian LPF (BT product = 0.3 in GSM) before the FM modulator to smooth frequency transitions.
• Lower BT: narrower spectrum, more ISI. Higher BT: less ISI, wider spectrum. BT = infinity recovers standard MSK.
• Constant envelope property is preserved — amplitude never varies, enabling nonlinear power amplifiers.
• GSM uses GMSK with BT = 0.3 at 270.833 kbps in 200 kHz channels, with Viterbi equalization to handle ISI.
• Modulation index h = 0.5 is the same as MSK — GMSK is MSK with Gaussian pulse shaping, not a different modulation index.

Quick Revision

• GMSK = MSK + Gaussian LPF prefilter. Modulation index h = 0.5 (unchanged from MSK).
• BT product controls filter bandwidth: lower BT = narrower spectrum but higher ISI.
• GSM uses BT = 0.3, bit rate = 270.833 kbps, channel spacing = 200 kHz.
• Constant envelope maintained — compatible with nonlinear Class C amplifiers.
• ISI from Gaussian filtering is handled by the Viterbi equalizer in GSM receivers.
• Exam trap: GMSK does NOT change the modulation index. BT = 0.3 refers to the Gaussian filter, not h.
• Gaussian filter formula: h(t) = C * B * exp(-2π^2 B^2 t^2 / ln2), where B is 3-dB bandwidth.