A.3 Performance Objectives
26.2673GPPeCall data transferGeneral descriptionIn-band modem solutionRelease 17TS
Performance Objective 1: The overall average transmission time should be as small as possible.
Performance Objective 2: Under all test conditions, a candidate should be as good as or better than the proposed eCall_via_CTM* (see 3GPP TR 26.967 [4]) would be.
NOTE: The objectives in the present document are intended as guidelines for designers of eCall solution candidates. The exact rules of candidate selection are specified in a separate document (PD3, "eCall Selection Test"). Objective 1 will be considered in the formulation of these rules. Objective 2 is intended only as a guideline and will not be considered in the formulation of the selection rules.
Performance Objective 1 is explained in detail in the following:
For any particular test condition (specified by speech codec plus radio channel error condition), the observed transmission time of the 140 bytes of the MSD may vary depending on the parameters of the channel simulation and the specific contents of the MSD. Therefore each MSD transmission may be regarded as one trial k in a random experiment, where the observed transmission time, Tk, is the random variable of interest. For each particular test condition C, the MSD transmission is repeated with different, randomly generated MSD data for at least 100 times (k = 1, 2, …, n, where n100) to get enough statistical significance.
To ensure a practical limit on the time required for testing a candidate, the observed value of Tk must have a reasonable upper bound. This upper bound, tUB, is fixed at a value of 200 seconds for one trial for all test conditions. Any value of Tk that is observed to be greater than tUB will be classified as a transmission failure and will be assigned the value of tUB.
Each particular test condition C gives an observed sample distribution T1, T2, …, Tn. The statistic of interest is the average value, C = (T1 + T2 + … + Tn) / n.
The Figure of Merit (FoM) over all test conditions is calculated by unweighted averaging of C over all particular test conditions C1, C2, …, Cm. A low Figure of Merit is – obviously – better than a higher Figure of Merit. The candidates will be ranked by their Figures of Merit.
The following assumptions are made for the measurement of Tk.
1. The starting time of the transmission with respect to speech codec audio frames is uniformly distributed.
2. The channel error condition is modelled by an error pattern obtained from offline simulations. The following radio conditions will be tested:
– GMSK Full Rate radio channel at C/I values of 1, 4, 7, 10, 13, 16 dB, and error free; with ideal frequency hopping, with the Typical Urban profile and with slow vehicle speed. These channel conditions will be applied in both directions (uplink and downlink) symmetrically.
– GMSK Full Rate radio channel at RSSI value -100 dBm with no other interferer. This channel condition will be applied in both directions (uplink and downlink) symmetrically.
The following speech Codecs will be tested: GSM_FR and FR_AMR (12.2, 10.2, 7.95, 7.4, 6.7, 5.9, 5.15, 4.75 kbps). DTX will be enabled in both directions.
Table A.1 gives an allocation of Codec Conditions to radio conditions in order to reduce the test effort to the reasonable minimum. The detailed allocation is defined in PD3, the "Selection Test Plan" Permanent Document.
Table A.1
|
GSM_FR |
12.2 |
10.2 |
7.95 |
7.4 |
6.7 |
5.9 |
5.15 |
4.75 |
|
|
C/I = 1 dB |
X |
||||||||
|
C/I = 4 dB |
X |
X |
X |
||||||
|
C/I = 7 dB |
X |
X |
X |
X |
X |
X |
X |
X |
X |
|
C/I = 10 dB |
X |
X |
X |
X |
X |
||||
|
C/I = 13 dB |
X |
X |
X |
||||||
|
C/I = 16 dB |
X |
||||||||
|
error free |
X |
X |
|||||||
|
RSSI = -100 dBm |
X |
X |
3. It is assumed that the transmission in the wireline part of the eCall uses PCM (G.711, A-law) without any further transcoding and with a fixed, pre-selected level setting.
4. It is assumed that no acoustical echo is produced by the IVS and that therefore no Acoustic Echo Suppressor is applied in the network.
5. It is assumed that no Hybrid Echo is produced by the PSAP connection and that therefore no Hybrid Echo canceller is applied in the network.
6. The MSD will contain randomly generated data. (Each possible byte sequence is considered to be equally probable.)
7. The round-trip delay between the IVS and PSAP is a randomly generated value in the range (200 ms, 220 ms).