Strategies for Handling High Peak Power in LTE Transmitters
Orthogonal frequency division multiple access (OFDMA) offers many advantages to LTE, notably the ability to deliver high data rates and perform well even in low quality channels. However, the technology does present some difficulties. In OFDMA systems the peak-to-average power ratio (PAPR) of composite signals can have peaks that exceed 12 dB above the average signal power. This complicates signal transmission by an e-Node B (eNB or base station) operating at high power. To avoid producing out-of-channel distortion products, eNB power amplifiers must have a high degree of linearity. However, power amplifiers with high linearity are expensive and modest in their power efficiency.
Steps can be taken to reduce the signal peaks and to extend the useful operating range of the power amplifier. Two complimentary methods are used for this purpose: crest factor reduction (CFR), which attempts to limit the signal peaks, and predistortion, which attempts to match the signal to the non-linear characteristics of the amplifier. Both methods are DSP-intensive, with predistortion being the more advanced method with the best performance but also the more difficult method to implement. For this reason CFR is usually the method used first.
Although it is necessary to control PAPR in LTE user equipment (UE), the use of SC-FDMA in the uplink rather than OFDMA means that the PAPR of the signal is no worse than that of the underlying modulation depth for QPSK or 16QAM. These are, respectively, 4 dB and 2 dB better at 0.01% probability than the Gaussian peaks typical of OFDMA. When requirements for 64QAM are introduced in the future, the UE design may then need to include CFR to limit the PAPR. (The more sophisticated predistortion technique is less suitable for use in the UE.) At this time, however, the strategies for handling peak power in LTE transmitters are focused on the eNB.
Crest factor reduction
CFR was first widely used with CDMA signals and is also an important technique for LTE, although the specifics of the implementation are somewhat different. CFR is distinct from pre-distortion in that it attempts to limit the peaks in the signal before it reaches the amplifier rather than shaping the input signal to compensate for amplifier nonlinearity. As such, CFR is a general technique that can be applied to any amplifier design. CFR improves headroom at the cost of degraded in-channel performance. OFDM signals without CFR have RF power characteristics similar to that of additive white Gaussian noise (AWGN), with peak power excursions more than 10 dB above the average power. It is impractical to design and operate power amplifiers with this level of headroom. Careful use of CFR can substantially reduce peak power requirements while maintaining acceptable signal quality.
All CFR techniques involve balancing PAPR improvements to alleviate out-of band distortion versus the detrimental impact on in-channel distortion and the cost and power associated with the increased baseband processing overhead. Perhaps the simplest type of CFR is clipping or signal compression, in which intermittent RF power peaks are either removed (by means of a simple clipping algorithm) or scaled (in the case of compression). Because of the dramatic effects of clipping and compression on signal quality and the availability of increased processing power, more advanced techniques are also used.
An example of a more sophisticated CFR technique is tone reservation, which is possible only on the downlink since the eNB can control the resource allocation for every resource block. A dummy transmission is allocated on the reserved tones, which cancels out the peaks generated by the composite signal, removing them from the useful tones. Although tone reservation promises to accomplish CFR without degrading signal quality, it has several disadvantages that need to be considered, including a loss of spectral efficiency (due to the loss of some subcarriers), loss of useful power, and increased computational overhead.
The effectiveness of CFR can be evaluated using the complementary cumulative distribution function (CCDF) applied to a series of instantaneous power measurements. (See sidebar.) The CCDF can be measured after the CFR operation and compared with either the actual or predicted CCDF of the signal without CFR. The CCDF measurement will yield a family of direct measurements of the reduction in PAPR, along with an associated probability of a specific PAPR. Measuring the power CCDF of a signal can be misleading if the signal is not stable; for instance, if some or all of the signal is not continuously transmitted. In such cases, CCDF measurements should be made time-specific.
A benefit of applying CFR is to increase the headroom available within the amplifier. This headroom can either be used to drive the amplifier harder or to improve the out-of-channel performance as measured by ACLR or SEM, or by some combination of the two. To choose the correct operating point for CFR, it is essential also to measure the in-channel quality (EVM, etc.) since in all cases, this quality will be degraded.
Predistortion enables the use of amplifier technologies that are both more power-efficient and less costly, although predistortion also adds design and operational complexity. Predistortion is a more advanced power management technique than CFR because it requires tight coupling to a specific amplifier design. Predistortion maintains the in-channel performance while operating in the non-linear region of the amplifier. This minimizes signal compression so that out-of-channel performance does not degrade at the higher operating level. A number of analog and digital predistortion techniques are available, from analog predistortion to feed-forward techniques and full adaptive digital predistortion. These techniques operate with varying degrees of effectiveness over varying bandwidths, and the eNB design may use a combination of techniques including predistortion and CFR to optimize overall cost and performance.
While CFR and predistortion are valuable techniques for managing high peak power in amplifiers, they can affect the signal and add complexity to the interpretation of RF measurement results. Therefore their use in LTE transmitters must be carefully evaluated versus the signal quality.
Sidebar: CCDF Measurement
CCDF curves provide critical information about the signals encountered in broadband systems. These curves also provide meaningful peak-to-average power data needed to describe the stress on a communication system. The two diagrams shown here illustrate this relationship graphically, mapping the time domain of the waveform to the CCDF curve. The x axis shows the signal power in dB above the root mean square (rms) value. The y axis shows the percentage of time that the signal spends at or above that level.
About the Authors
Ben Zarlingo is a product manager for communications test with Agilent Technologies' Signal Analysis Division. He received a BS in Electrical Engineering from Colorado State University in 1980 and has worked for Hewlett-Packard /Agilent Technologies in the areas of spectrum, network and vector signal analysis, with a primary focus on techniques for the design and troubleshooting of emerging communications technologies.
Moray Rumney joined Hewlett-Packard/Agilent Technologies in 1984, after completing a BSc in Electronics from Edinburgh’s Heriot-Watt University. Since then, Moray has enjoyed a varied career path, spanning manufacturing engineering, product development, applications engineering, and most recently technical marketing. His main focus has been the development and system design of base station emulators used in the development and testing of cellular phones. Moray joined ETSI in 1991 and 3GPP in 1999 where he was a significant contributor to the development of type approval tests for GSM and UMTS. He currently represents Agilent at RAN WG4, developing the air interface for HSPA+ and LTE. Moray has published many technical articles in the field of cellular communications and is a regular speaker and chairman at industry conferences. He is a member of IET and a chartered engineer.