Introduction to GSM

Figure 1:  Number of cellular phone subscribers worldwide (actual and estimated)

In the near future, UMTS (Universal Mobile telephone system) will be launched. The new communication standard will feature real-time video, handled by an high-speed protocol that will not be compatible with the GSM standard and existing infrastructure (Figure 2).

Figure 2:  Cellular phone development from the GSM standard to the 3G UMTS standard

Figure 3:  Frequency assignment in GSM and DCS bands

The handset in the GSM standard is called the mobile station (MS). The base station is referred to as BS. The maximum GSM emission power is 2W in 900MHz and 1W in 1800MHz. The base station may reach 300W, but many efforts are undergoing to avoid the use of such power in urban environments, as base stations and mobile phones may harm health. Figure 3 describes the frequency allocation for up and down links. A significant separation of the Tx and Rx sub-bands eases the coupling avoidance between transmitter and receiver parts.

Figure 4:  Network channel capacity and separation between operators

The theoretical network capacity is the number of available channels in the sub-band. As each frequency channel is 200KHz in size, up to 125 channels are available for GSM and 375 channels for DCS (Figure 4). This number seams quite large, however, operators buy licenses to exclusively use a portion of the channels. For example, Operator 1 in Figure 4 exploits the frequency band from 1710 to 1735MHz, 125 channels in DCS.

Figure 5:  Sharing the frequency resource between adjacent cells

A further reduction of channels is due to the cellular structure of the network (Figure 5). To avoid frequency conflicts, the channel resource is split into seven portions, each one allocated to a cell. Consequently, the number of available channels in our example is 125/7, which equals 17 available channels.

Figure 6:  An illustration of time domain multiplexing

Figure 7:  Block diagram of a mobile station

The block diagram in Figure 7 provides a simplified description of the mobile station. A microphone captures the sound, which is sampled in a numerical format, compressed, coded, and modulated. A high-frequency oscillator translates the modulated signal to a valid transmission frequency. The received signal (less than 1mV) is amplified before down-conversion to a low-frequency, demodulation, decoding, and sound reconstruction.

The sound captured by the microphone is filtered to remove harmonics lower than 300Hz and higher than 3000Hz (Figure 8). A gain-control stage then keeps the signal envelope more or less constant, before a sampling at the rate of 8000Hz, in a 13-bit format. This leads to a rate of 104 Kbit/s, meaning that a compression algorithm is mandatory to obtain an acceptable rate of 12Kb/s.

Figure 9:  The GSM speech coder is based on adaptive filtering techniques

The voice coder implemented in the GSM handset splits the sound into portions of 20ms, with a time-domain aspect reported in the top portion of Figure 9. The coder removes the redundant periodic structure of the sound (80% of the speech). You use linear prediction and root mean square minimization as the mathematical basis for computing the best coefficients to approximate each portion of sound by an adaptive filter. Consequently, the filter coefficients are transmitted, not the sampled sound. This features a significant reduction of data. Low-order coefficients are coded in a 6-bit format while high-order coefficients are in a 3-bit format.

Figure 10:  The GSM modulation, based on phase change

Figure 11:  The GSM demodulation obtained using direct down-conversion

Figure 12:  Byte reconstruction using cosine and sinus multiplication