Figure 7 shows the architecture of protocols used within the GSM system, with signalling protocols, interfaces as well as the entities already shown in Figure 5.
Again the main area of focus is in the Um interface, this is because the other interfaces occur between entities in a fixed network. The physical layer, Layer 1 handles all the radio specific functions. This layer includes the creation of bursts according to the five different formats, the multiplexing of bursts into TDMA frames, synchronisation with the BTS, detection of the idle channels and the measurement of the channel quality on the downlink. At Um, the physical layer uses GSMK (Gaussian Shift Minimum Keying) for the digital modulation and performs encryption/decryption of data This means that encryption is not performed end-to-end, but only between MS and BTS over the air interface.
The synchronisation also includes the correction of the individual path delay between the MS and the BTS, all MSs within a cell can use the same BTS and hence must be synchronised to the BTS. This is due to the fact that the BTS generated the time-structure of the frames and slots etc. This can be problematic since in this context there are different RTTs (Round Trip Time). An MS that is close to the BTS has a very short RTT whereas an MS that is 35 km away has a RTT of around 0.23 ms.. If the MS 35 km away used the slot structure without correction, a large guard spaces would be required as 0.23 ms. are already 40% of the 0.577 ms available for each time slot. (Wray Castle, GSM Appreciation, 1998.). Therefore the BTS sends the current RTT to MS, which then adjusts its access time so that all bursts reach the BTS within their limits. This mechanism ensures that the guard space is reduced to only 30.5 ms or 5%. See Figure 7. This means the adjustment of the access is controlled via the variable timing advance, where a burst can be shifted up to 63 bit times earlier, with the resulting bits having a duration of 3.69 ms, thus will result in the 0.23 ms needed.
The physical layer has several main tasks that comprise the channel coding, error detection/correction; this is directly combined with the coding mechanisms. FEC (Forward Error Correction) is used extensively in the coding channel, FEC adds redundancy to the user data, thus allowing for the detection and correction of selected errors. The power of the FEC scheme depends on the amount of redundancy, coding algorithm, and any further interleaving of data to minimise the effects of burst errors. Whatsmore the FEC is the reason that error detection/correction occurs in the physical layer. This differs to the ISO/OSI reference model where it occurs in layer two. The GSM physical layer tries to correct errors, however it does not deliver erroneous data to the higher layers.
GSM logical channels use different coding schemes with different correction capabilities, for example speech channels need the additional coding of voice data after analogue to digital conversion. This is in order to reach a data rate of 22.8 kbit/s (using the 13 kbit/s from the voice codec plus redundancy, CRC bits, and interleaving (Goodman, 1997). When GSM was envisaged it was assumed that voice would be the main service so the physical also contains special functions, for instance VAD (Voice Activity Detection), which transmits voice data only when there is a voice signal. In the duration between voice activity, the physical layer generates a comfort noise to fake a connection, however no actual transmission takes place.
All the interleaving in the voice channel is to minimise interference due to burst errors and the recurrence pattern of a logical channel generates a delay for transmission, although this delay is only about 60 ms for TCH/FS and about 100 ms for TCH/F9.6. These times have to be added to the transmission delay if the BTS is communicating with an MS rather than a standard fixed station (for example a stationary computer etc.) and this in turn may influence the performance of any of higher layer protocols, e.g.. for computer data transmission.
Signalling between the entities within the GSM network requires the use of the higher layers (see Figure 7). For this, the LAPDm (Link Access Procedure for the D-Channel) protocol has been defined at the Um interface for layer two. As the name already implies, it has been derived from link access procedure for the D-Channel (LAPD) in the ISDN system, which is a version of HDLC (Goodman, 1997), LAPDm is a lightweight version of LAPD, in that it does not require synchronisation flags or check summing for error detection, these are not needed as these functions are already performed in the physical layer of the GSM network. LAPDm, however offers reliable data transfer over connections, re-sequencing of data frames and flow control (ETSI, 1993, ETS 300 937), (ETSI, 1999) TS 100 938. Due to the fact that there is no buffering between layer one and two, the LAPDm has to obey the frame structures, recurrence patterns etc defined for the reassembly of data and acknowledged/unacknowledged data transfer.
Layer three in the GSM network is made up of several sublayers as shown in Figure 7, the lowest sublayer is the RR (Radio Resource Management). Only part of this layer the RR’, is implemented in the BTS, the remainder of the RR is situated in the BSC. The BSC via the BTSM (Base Transceiver Station Management) are responsible for the functions of the RR’. The RR’ has the function of setting up, maintenance and release of the radio channels. Also the RR’ has direct access to the physical layer for radio information and offers a reliable connection to next higher layer.
The physical layer has several main tasks that comprise the channel coding, error detection/correction; this is directly combined with the coding mechanisms. FEC (Forward Error Correction) is used extensively in the coding channel, FEC adds redundancy to the user data, thus allowing for the detection and correction of selected errors. The power of the FEC scheme depends on the amount of redundancy, coding algorithm, and any further interleaving of data to minimise the effects of burst errors. Whatsmore the FEC is the reason that error detection/correction occurs in the physical layer. This differs to the ISO/OSI reference model where it occurs in layer two. The GSM physical layer tries to correct errors, however it does not deliver erroneous data to the higher layers.
GSM logical channels use different coding schemes with different correction capabilities, for example speech channels need the additional coding of voice data after analogue to digital conversion. This is in order to reach a data rate of 22.8 kbit/s (using the 13 kbit/s from the voice codec plus redundancy, CRC bits, and interleaving (Goodman, 1997). When GSM was envisaged it was assumed that voice would be the main service so the physical also contains special functions, for instance VAD (Voice Activity Detection), which transmits voice data only when there is a voice signal. In the duration between voice activity, the physical layer generates a comfort noise to fake a connection, however no actual transmission takes place.
All the interleaving in the voice channel is to minimise interference due to burst errors and the recurrence pattern of a logical channel generates a delay for transmission, although this delay is only about 60 ms for TCH/FS and about 100 ms for TCH/F9.6. These times have to be added to the transmission delay if the BTS is communicating with an MS rather than a standard fixed station (for example a stationary computer etc.) and this in turn may influence the performance of any of higher layer protocols, e.g.. for computer data transmission.
Signalling between the entities within the GSM network requires the use of the higher layers (see Figure 7). For this, the LAPDm (Link Access Procedure for the D-Channel) protocol has been defined at the Um interface for layer two. As the name already implies, it has been derived from link access procedure for the D-Channel (LAPD) in the ISDN system, which is a version of HDLC (Goodman, 1997), LAPDm is a lightweight version of LAPD, in that it does not require synchronisation flags or check summing for error detection, these are not needed as these functions are already performed in the physical layer of the GSM network. LAPDm, however offers reliable data transfer over connections, re-sequencing of data frames and flow control (ETSI, 1993, ETS 300 937), (ETSI, 1999) TS 100 938. Due to the fact that there is no buffering between layer one and two, the LAPDm has to obey the frame structures, recurrence patterns etc defined for the reassembly of data and acknowledged/unacknowledged data transfer.
Layer three in the GSM network is made up of several sublayers as shown in Figure 7, the lowest sublayer is the RR (Radio Resource Management). Only part of this layer the RR’, is implemented in the BTS, the remainder of the RR is situated in the BSC. The BSC via the BTSM (Base Transceiver Station Management) are responsible for the functions of the RR’. The RR’ has the function of setting up, maintenance and release of the radio channels. Also the RR’ has direct access to the physical layer for radio information and offers a reliable connection to next higher layer.