The transport channel usage ratio is a simple formula that describes the relation between maximum possible data rates provided by transport channels and the average throughput measured when a channel has been active. This time is called the observation period in the transport channel usage ratio formula shown below:
DataVolume of RLC TransportBlocksðkbitị
max Theoretical transport channel throughputðkbit=sị observation periodðsị100%
ð2:9ị Figure 2.22 TCP and FTP data frame
Figure 2.23 UDP datagram with length indicator
Using this formula a good ratio can be computed that describes how often (in %) it has been necessary to use the maximum bandwidth of a single transport channel. Note that this analysis must be done separately for uplink and downlink traffic.
For a single call analysis it might be enough to know the transport channel ID. Usually the same transport channel ID is used to identify a DCH even if the call was in CELL_FACH state for a period of time and all dedicated transport channels have been deleted. However, if the analysis of multiple connections need to be performed on a certain aggregation level such as cell or SRNC it is necessary to know the radio bearer type of the DCHs (uplink/
downlink maximum bit rate on the radio interface following definitions of 3GPP 34.108).
This is because the same channels having the same DCH-ID may have different transport format settings and only measurement results from channels of the same type/transport format can be used to calculate an average of multiple measurements. Here it must also be kept in mind that this radio bearer type assigned to a certain DCH by the call admission control function of SRNC may change during the call due to dynamical reconfiguration as described in section 1.1.6.
Table 2.19 shows an example of such an analysis for a single cell. It makes sense to aggregate measurements of multiple calls using UL/DL combinations of maximum bit rates to gain a better overview of how single connections of the same type behave. A low ratio value for a certain maximum data rate does not necessarily indicate a problem, it just means that not much data has been transmitted on these UL DCHs, for example because the used application service is a downlink file transfer (in this case only TCP Acknowledgements and RLC AM Status PDUs over a long time on UL are recognised). Another aspect is that on uplink a low transport channel usage ratio must not be seen as critical, because the number of uplink channelisation codes is not a critical factor, as each uplink dedicated physical channel is uniquely identified by an uplink scrambling code. Channelisation codes on uplink are only necessary to distinguish between DPCCH and DPDCH of the same connection. The situation on downlink is different. Here a single cell uses the same downlink scrambling code for all connections, hence the channelisation code is a unique identifier of a single connection. But the number of available channelisation codes per cell is limited due to the structure of the channelisation code tree. Also the number of available codes determines the capacity limit of the cell. To give an example, there are only four channelisation codes with spreading factor 4, which represents the highest downlink data rate possible for a single connection on the radio interface in non-HSDPA capable cells. If these four available spreading codes are used simultaneously no other connections are possible using the same cell. The spreading factor is indirectly (using the coding rate of the convolutional coder) related to the downlink maximum bit rate of the radio bearer, which represents the maximum
Table 2.19 Example of transport channel usage ratio analysis in a certain SRNC UL transport channel DL transport channel
usage ratio usage ratio Cell 1
RB-Type UL/DL 64/64 kbps 60% 85%
RB-Type UL/DL 64/128 kbps 25% 80%
RB-Type UL/DL 64/384 kbps 25% 90%
theoretical transport channel throughput used to compute the transport channel usage ratio.
If the transport channel usage ratio of channels with the highest available downlink bit rate is low, this means that these transport resources are not efficiently administrated by the SRNC and high-speed transport resources are blocked by users that need much lower bit rates, but for lower bit rates higher spreading factors are used. And the higher the spreading factor the more connections can be served in a cell at the same downlink bit rate.
If a cell is HSDPA capable the transport channel usage ratio of HS-DSCH cannot be computed, because it is very difficult to estimate the theoretical maximum throughput of an HS-DSCH. Large parts of data volume transported on this channel are hidden when monitor- ing the Iub interface. This is because retransmissions on the radio interface due to hybrid ARQ (HARQ) error detection/correction cannot be monitored on lub, but will most likely require a huge portion of HS-DSCH transport capabilities.
If the data volume of RLC transport blocks is counted, analysis algorithms run again into a problem discussed in the section about transport channel throughput. As it was pointed out in this section an RLC transport block contains not only the payload to be transported, but also RLC/MAC header bits. If the transport channel throughput is measured related to a 64 kbps radio bearer using full transport block size of 336 bits, it must also calculate the maximum theoretical transport channel throughput using 336 bits. To base the measurement on a transport block size of 336 bits leads to a maximum value of 67.2 kbps for a 64 kbsp radio bearer. The advantage of this variant of measurement is that it reflects the real traffic situation on the DCH, which is also used to transport the RLC header and padding bits as well as retransmitted RLC frames and RLC signalling messages (status PDUs). The resulting volume of all these different kinds of data finally also determines which spreading factor needs to be chosen for radio interface transmission.
Some radio network planners have a different point of view of this KPI and say it must be based on the data volume of RLC SDUs instead of the data volume of RLC transport blocks as shown in Equation (2.10):
DataVolume of RLC SDUs ðkbitị
max Theoretical transport channel throughput ðkbit=sị observation periodðsị100%
ð2:10ị An explanation of this requirement is that the target of transport channel usage ratio analysis is to find out how well the RNC admission control function is able to assign the necessary transport resources for a particular payload bit stream. If the data volume of RLC SDUs is in the numerator the maximum theoretical transport channel throughput must be computed in a different way. Instead of the size of transport blocks it must be based on the available payload size of transport blocks. For transport channel throughput measurements it has already been demonstrated that the payload size of transport blocks is not a fixed value as assumed in 3GPP 34.108, but for a 336-bit transport block the payload size varies for single transport blocks between 38 bytes (304 bits) and 40 bytes (320 bits). If the maximum theoretical transport channel throughput is calculated using 320 bits, a maximum is defined that can never be reached. Hence, the maximum ratio calculated using this approach is not at 100%, but instead at 98 or 99%, which needs to be taken into account as a natural measurement tolerance. If the admission control function in SRNC software really uses user- perceived throughput (because this is what the data volume of SDUs stands for) to assign
necessary radio resources to a connection is a question that can only be answered by network equipment manufacturers (NEMs) and this answer – if given at all – will always be a pro- prietary one.
Finally there is another aspect that may cause tolerance in measurement results in the range of up to approximately 10%: it is difficult to agree on a common definition when a transport channel is really available to transport data. Some measurement manufacturers use the ALCAP Establish Confirm message, which acknowledges the set up of an appropriate lub/Iur physical transport bearer (AAL2 SVC); others wait until a number of FP synchro- nisation frames have been sent on the uplink/downlink of this AAL2 connection. However, there is no signalling message that indicates that the synchronisation process is finished. And in addition FP synchronisation frames are also sent as a check during long periods of DCH activity. A third option is to define the first user plane frame seen on AAL2 SVC as the start of availability, but in this case it is possible that in two directions (uplink or downlink) the first DCH frame will be seen quite late if there is no initial data to be transmitted in this direction. The time difference between ALCAP Establish Confirm and the first DCH frame is usually not more than a few hundred milliseconds. However, for PS connections with multiple channel type switching procedures these tolerance time frames also need to be multiplied, which leads to the fairly high measurement tolerance as mentioned at the beginning of this paragraph.