21.3.1.1 Signal Quality: Error Vector Magnitude (EVM)
The quality of the transmitted radio signal has to fulfil certain requirements. The main parameter used to measure this quality is the Error Vector Magnitude (EVM), which is a
Channel Bandwidth [MHz]
Transmission Bandwidth Configuration [RB]
Transmission Bandwidth
Active Resource Blocks
in downlink
Resourceblock Channel edge
Channel edge
d.c. in baseband) is not transmitted Centre subcarrier (corresponding to
measure of the distortion introduced by the RF imperfections of practical implementations.2 It is defined as the magnitude of the difference between a theoretical reference signal (i.e.
the signal defined by the physical layer specification equations) and the actual transmitted signal (normalized by the intended signal magnitude). The EVM sets the maximum possible SNR of a radio link in the absence of any noise, interference, propagation loss and other distortions introduced by the radio channel; it therefore serves to determine the maximum useful modulation order and code rate.
The EVM measures the quality of the transmitted signal across all the allocated RBs.
The measurement duration is one slot for the UE (uplink) and one subframe for eNodeB (downlink) and takes into account all the symbols belonging to the modulation scheme under test. Figures 21.2 and 21.3 show the EVM measurement points in the downlink and uplink transmission chains respectively [5, 6].
The EVM measurement is taken after an equalizer in the test equipment, which carries out per-subcarrier channel correction [5, 6]. The equalizer is used in order to obtain a measurement which realistically shows what a receiver might experience. It is intended to reflect the fact that the equalizer in the receiver is capable of correcting some of the impairments of the transmitted signal to some extent. A zero-forcing equalizer is used for EVM measurement;3however, real receiver implementations may use different equalization techniques, and therefore the measured EVM may not exactly correspond to the actual signal quality experienced by all receivers.
<3J)0JJ
4JJJJ :JJ ,,)J
@JJ *& 3"J : KJ
::1J
J JJ*DJ J )K"J
synchronization
Figure 21.2: Measurement points for the EVM for the downlink signal.
2Note that distortion may also be introduced in the process of balancing in-channel and out-of-channel performance, especially in the eNodeB.
3Details of how to compute the coefficients of the equalizer can be found in [5, 6].
>*)0
4
: ,,)
@
*& 6-,)
*D
6:
Figure 21.3: Measurement points for the EVM for the uplink signal.
It should be noted that before measuring the EVM, time and frequency synchronization must be carried out. The test equipment then computes the EVM for two extreme values of sample timing difference between the FFT processing window and the nominal timing of the ideal signal; these two extreme values correspond to the beginning and end of the window, the length of which is expressed as a percentage of the Cyclic Prefix (CP) length. Finally the measurement is averaged over 20 slots for UE (uplink) and 10 subframes for eNodeB (downlink) to reduce the impact of noise.
In LTE the EVM is required to be less than 17.5% for QPSK, 12.5% for 16QAM and 8%
for 64QAM (64QAM being applicable to the downlink only) [5, 6]. These EVM values are designed to correspond to no more than a 5% loss in average and cell-edge throughputs in typical deployment scenarios. At the link level, EVM is equivalent to an SNR loss.
A description of the main transmitter impairments which generally give rise to non-zero EVM is given in Section 21.5.1.
21.3.1.2 Transmit Output Power
The transmitted output power directly influences the inter-cell interference experienced by neighbouring cells using the same channel, as well as the magnitude of unwanted emissions outside the transmission band. This affects the ability of the LTE system to maximize spectral efficiency, and it is therefore important that the transmitters can set their output power accurately.
For the eNodeB, the maximum output power must remain within±2 dB of the rated power, (PRAT) declared by the manufacturer. There are three classes of eNodeB (see Section 24.4):
the Home eNodeB withPRAT≤20 dBm, the Local Area eNodeB withPRAT≤24 dBm and the Wide Area eNodeB for which no upper limit has been defined. In addition, the dynamic range in the frequency domain (computed as the difference between the power of a given
Resource Element (RE) and the average RE power) must not exceed specified limits [5, 6]
depending on the modulation order, in order to avoid saturating the UE receivers.4
For the UE, a requirement is defined for only one power class, known as ‘power class 3’
for which the maximum output power is 23 dBm. Like the eNodeB, the UE must satisfy this requirement within a range of±2 dB.
The UE maximum output power requirements may be modified by a number of factors [5]:
• Maximum Power Reduction (MPR). The purpose of MPR is to allow the UE, in some demanding configurations, to lower its maximum output power in order to meet the general requirements on signal quality and Out-Of-Band (OOB) emissions. Note that the MPR is an allowance and the UE does not have to use it. Table 21.5 shows the allowed values of MPR as a function of the modulation scheme, the channel bandwidth and the transmission bandwidth (number of transmitted RBs).
Table 21.5: Maximum power reduction for power class 3.
Modulation Channel bandwidth (MHz)/ MPR (dB) transmission bandwidth (RB)
1.4 3 5 10 15 20
QPSK >5 >4 >8 >12 >16 >18 ≤1
16QAM ≤5 ≤4 ≤8 ≤12 ≤16 ≤18 ≤1
16QAM >5 >4 >8 >12 >16 >18 ≤2
• Additional MPR (A-MPR). The eNodeB may inform the UE of the possibility of further lowering its maximum power by signalling an A-MPR. The need for A- MPR occurs with certain combinations of E-UTRA bands, channel bandwidths and transmission bandwidths for which the UE must meet additional (more stringent) requirements for spectrum emission mask and spurious emissions (see Section 21.3.2).
As with MPR, the A-MPR is an allowance, not a requirement, and it applies in addition to MPR. Regardless of whether the UE makes use of the allowed MPR and A-MPR, the additional requirements for spectrum emission mask and spurious emissions that are signalled by the network always apply. The reason for the complex set of conditions for the relaxation is the expected intermodulation products which may fall into adjacent bands which have different levels of sensitivity (e.g. public safety bands).
• ΔTC.ΔTCis a 1.5 dB reduction in the lower limit of the maximum output power range when the signal is within 4 MHz of the channel edge.
• Power Management MPR (P-MPR). Introduced in Release 10, the P-MPR allows a UE to reduce its maximum output power when other constraints are present. In particular, multi-RAT5 terminals may have to limit the LTE transmission power if
4For example, a Physical Downlink Control CHannel (PDCCH) RE power can range between−6 dB and+4 dB around its average.
5Radio Access Technology.
transmissions on another RAT are taking place simultaneously. Such power restrictions may arise, for example, from regulations on Specific Absorption Rate (SAR) of radio energy into a user’s body or from out-of-band emission requirements that may be affected by the inter-modulation products of the simultaneous radio transmissions.
The P-MPR is not aggregated with MPR or A-MPR, since any reduction in a UE’s maximum output power for the latter factors helps to satisfy the requirements that would have necessitated P-MPR.
Taking these factors into account, the UE has to configure its nominal maximum power PCMAX(i.e. the highest power at which the UE will attempt to transmit) within the following upper and lower limits:
PCMAX, L=min (PEMAX−ΔTC,PPowerClass−max (MPR+A-MPR,P-MPR)−ΔTC)
PCMAX, H=min (PEMAX,PPowerClass) (21.1)
wherePPowerClassis the original power class of the UE andPEMAXis a maximum power that may be signalled by the network.
Note that when carrier aggregation is configured (for UEs of Release 10 and beyond), PCMAX becomes a component-carrier-specific nominal maximum power, PCMAX,c, as explained in Section 28.3.5.
Finally, the actual transmitted power is allowed to vary over a wider range due to uncertainties in the transmit chain. The added tolerance is a function ofPCMAXwhich varies from 2 dB at high powers to 7 dB for powers below 8 dBm.
In summary, the concept of maximum power for the UE is a complex and dynamic function of many variables designed so that the needs of specific operating conditions can be met without over-specifying the transmitter design. Details of the maximum power specification can be found in [5].
21.3.1.3 Output Power Dynamics
LTE, like UMTS and HSPA, needs to ensure user orthogonality in the time domain. In WCDMA, the requirements for this were relatively straightforward being based on so-called
‘on–offmasks’ which define the allowable transmit power during the ‘off’ and ‘on’ periods of transmission. For LTE, similar general requirements exist for the eNodeB in both FDD and TDD [6]. However, for the LTE UE the requirements are considerably more complex than those for WCDMA due to the characteristics of the SC-FDMA scheme used for the uplink transmissions. Figure 21.4 shows the general ‘on/offtime mask’ which applies whenever the UE is required to switch on or off.6Note that although the requirement is given in terms of a
‘mask’, it actually applies to the average power during the ‘on’ and ‘off’ periods [5].
In UMTS the transient period (20μs) is centred on the slot boundary, while in LTE the transient period is in general shifted forwards into the next subframe. Hence, the first few SC- FDMA samples are vulnerable to being corrupted by insufficient transmit power or inter-UE interference whereas the samples at the end of the subframe are protected. For the Physical Random Access CHannel (PRACH) the transient periods are located outside the preamble in order to protect the whole PRACH preamble. The same applies to Sounding Reference
6This is also applicable for Discontinuous Transmission (DTX) measurement gaps – see Chapter 22.
Figure 21.4: General on–offtime mask for the UE. Reproduced by permission of©3GPP.
Signals (SRSs) (see Section 15.6) which need to be protected to ensure the eNodeB can exploit them entirely7 to carry out reliable uplink channel sounding without impairments arising from transient periods. For TDD, where two SRSs can be transmitted on adjacent symbols in the UpPTS field, the transient periods are located between the two SRSs.
The general time masks apply in the case of transitions into and out of the offpower state.
A general output power dynamic requirement also exists for continuous transmission at slot boundaries where frequency hopping occurs and at subframe boundaries for either frequency hopping or power changes. In all these cases, the transient periods are symmetric about the slot or subframe boundaries. Transient power masks also apply in the following cases:
• Transition from PUCCH/PUSCH to SRS with DTX after SRS;
• Transition from PUCCH/PUSCH to SRS to PUCCH/PUSCH;
• Transition from DTX to SRS to PUSCH/PUCCH;
• Transition from PUCCH/PUSCH to DTX in an SRS symbol to PUCCH/PUSCH.