Green Solutions Offer Big Paybacks for Wireless Telecommunications

By Dane Elliot, Director of Marketing, Optichron, Inc.

Leading wireless telecommunications suppliers have announced strategic efforts to provide more efficient solutions for operators around the world, which can reduce basestation power consumption by significant amounts. These so-called “Green Solutions” use optimized hardware design, with innovative power amplifier (PA) and power consumption management to help operators realize a high level of power savings.  In particular, these solutions leverage new digital nonlinear signal processing approaches to deliver both capital and operational savings.

Why Does It Matter?

Reducing the power consumption of basestations has emerged as one of the key concerns of telecom operators worldwide. One might ask why now; is it a reflection of the Green directions spawned by awareness of global warming?  Is there a concern about where power will come from tomorrow, or is it possible that the service providers want to climb onto the positive PR bandwagon associated with Green?  While all of these motivations might be part of the drive to Green network operations, one immediate reason is actually pure profit.  In the operational environment, once the site is in place, the largest single expense is power.  Reducing power can have a significant benefit to the bottom line.

Where is the Gain?

These Green Solutions adopt leading PA and signal processing technologies that when combined can boost the power efficiency of basestations by 50% or more, sharply reducing the overall cost and power consumption of the facilities.  These new technologies also allow a single basestation to provide more bandwidth, which results in fewer, smaller basestations, providing more capacity at lower capital costs and operational expenses.  This increased bandwidth can cost less when traditional methods give way to more intelligent solutions with fewer expensive RF components.  Further, reducing the weight and size of a basestation allows it to be located in places that were previously prohibitive.  Smaller, lighter basestations also reduce the cost of the installations and make servicing them easier and less expensive.

Where Does the Power Go?

Electrical power is consumed everywhere in the system, but proportionally the PA is the dominant power consumer. When all power consumption is considered, the PA itself can easily consume as much as 75% of the operating power budget. The balance for electronics is used for the baseband processor, various digital processes, backhaul communications and the transceiver.  Finally, some of it can go to power air conditioning or fans to keep the system adequately cooled.  The important functional power is the RF power that is delivered through the antenna.  In some basestations, 20 W of RF power at the antenna can easily require over 200 W input to the system.

A Green Solution uses a distributed architecture that allows 20 W basestations to have the same output spectrum coverage as traditional 40 W basestations, and high-efficiency PA technologies that make possible convection cooling, direct cooling, and intelligent cooling implementations. An added benefit of reducing the need for powered fans and air conditioners is that these are also sources of noise pollution.   The ideal Green Solution can also be integrated with more environmentally friendly energy sources such as wind, solar power, and methane.

Meeting Protocol Specifications

Like most engineering challenges, managing the development of a basestation – Green or otherwise – is a continuous tradeoff.  Every attempt to improve performance in one area is balanced against a loss of performance in another area.  Our primary objective in developing a Green Solution is to minimize the operational costs of the installed product.  As described, an analysis of where power is consumed rapidly converges on the PA.  Where integration of digital functionality and process technology has had dramatic effects on power consumption in digital parts of the basestation, these techniques do not directly apply to the analog RF or PA. 

To make matters worse, as newer protocols are implemented that increase data rates through wider channel bandwidths and more complex modulation schemes, they adversely affect PA efficiency, moving operational costs in the wrong direction.

Clearly, concentrating on making the PA more efficient is what matters most, however traditional approaches offer limited improvements to solving analog problems. From both power consumption and capital perspectives we should concentrate on new approaches that are deployed in the digital domain in order to achieve the largest efficiency gains in the PA.

Issues to Consider

Protocols

Early protocols such as GSM operate with only 200 KHz wide carriers and simple modulation techniques.  As newer protocols have been defined, however, the carrier bandwidth has grown two orders of magnitude.  CDMA at 1.25 MHz has grown to WCDMA at 5 MHz, to WiMAX at 10 MHz, and now to LTE with a carrier bandwidth of 20 MHz.  While reasonable efficiencies have been achieved for the narrower carriers, traditional design methodologies have not achieved efficiencies at the higher carrier bandwidth.

PA Architecture

PA architecture is also evolving, increasing operating bandwidth and efficiency but at the same time creating challenges in linearity.  Class AB amplifiers are giving way to Doherty amplifiers, and while this can improve efficiency, it has a detrimental effect on the transmit chain linearity, which blunts the full value of the new architecture.

Multiple PAs and Combiners

For capital economies, a traditional approach is to combine the output of several narrower PAs to deliver wider effective bandwidth and multi-carrier solutions.  However, while increasing bandwidth this approach does just the opposite for operational efficiencies.  Combining the output of four 5 MHz WCDMA carriers can cost half the broadcast power, or in other words double the cost to operate the PA.

Traditional Methods

As bandwidths increase, modulation schemes become more complex, and multi-carrier architectures become the norm, the signal becomes more complex and difficult to manage.  The most obvious attribute of these more complex signals is an increase in peak-to-average signal ratio (PAR).  Increasing PAR generally requires the designer to operate the PA at a lower level of efficiency.  This is because the peaks cannot be allowed to drive the PA into saturation, causing spurious emissions.

Back Off the PA

The PA is generally operated at the P_-1dB point minus PAR.  The P_-1dB point is the point at which the PA drops 1dB from the projected linear operating line and is assumed to be just before the PA goes into saturation.  As above, we assume that driving the PA into saturation will cause spurious emissions.  In the example below, a PA with a 44 dBm P_-1dB point broadcasting a signal with a PAR of 13 dB would operate at 44 minus 13, or 31 dBm.  This assures that the peak signals will not saturate the PA causing spurious emissions.  If the PAR of the signal were 6 dB, the operating point could be 38 dBm and still assure operation without saturating the PA and causing unacceptable emissions.  Backing off the PA may be a traditional solution but the operational cost can be very high.

Every 3dBm of power doubles the output power of the system, making PAR management in the digital domain a must-have technology. PAR reduction allows designers working with data-centric technologies to improve the efficiency of costly PAs by allowing an increase in the operating point while maintaining spectral compliance.  Crest factor reduction (CFR) capability, in figure above and described below, can sharply reduce PAR, allowing designers to operate the PA in a more efficient operating range and thus gaining valuable headroom to improve system power efficiency at a reduced cost.

Every basestation designer faces the issue of meeting emission standards.  These are described as error vector magnitude (EVM) and adjacent channel leakage ratio (ACLR) among others.  EVM is a measurement of the ability to distinguish one symbol from the next.  Simple modulation techniques allow large EVM values because there are few signals in a constellation.  The newer modulation processes have 16, 32, 64 and more symbols to distinguish between and therefore require much tighter EVM specifications to achieve acceptable bit error rates.

ACLR is a measurement of out-of-band emissions.  It is important because, in our limited spectrum, we cannot afford to interfere with a signal next to our own any more than we can tolerate that signal to interfere with our own.

Once again, but for a slightly different reason, operating the PA at a lower level of efficiency helps us to achieve these ever-increasing performance requirements.  PAs are notoriously nonlinear.  Designers go to great lengths to improve linearity, and at lower bandwidths have been reasonably successful.  However, at the wider bandwidths and with more sophisticated modulation, achieving linearity is less than stellar.  Still, traditional approaches such as operating the PA in a more linear region can meet required specifications, but again this sacrifices efficiencies and increases operational and capital costs.

Other traditional approaches help as well.  Power transistor manufacturers continue to make strides at the physics level and continuously deliver better devices with which to design.  Newer PA architectures such as the Doherty architecture also address efficiency issues.  However, all of these approaches focus in the analog, mixed-signal domain. With the strides in digital process technologies, addressing these issues in the digital domain now presents a paradigm shift for solving the problem a better way.

Digital Solutions to A