Browse Prior Art Database

Modified Power Ramp for Accelerated PreDistortion Training

IP.com Disclosure Number: IPCOM000171771D
Publication Date: 2008-Jun-18
Document File: 4 page(s) / 175K

Publishing Venue

The IP.com Prior Art Database

Abstract

The high Peak to Average Ratio (PAR) of current and future wireless modulation standards imposes severe trade offs in the form of linearity versus efficiency for the transmitter. Adaptive PreDistortion (APD) is often adopted as a solution to this trade off problem. However, standards such as 3GPP, do not explicitly facilitate training of the APD Look Up Table (LUT) which can compromise its performance from a 'blind start' and following changes in the transmitter nonlinear characteristics due to, for example, a transient load change. The following publication details how the relatively less prescriptive section of the 3GPP standard can be exploited to insert dedicated training waveforms to accelerate the training of the APD LUT. In particular, the inter-slot boundary, power transient AM ramping waveform can be modified to improve training near the lower and upper enters of the LUT which otherwise would receive poorer training if the natural modulation waveform was solely rely on.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 52% of the total text.

Modified Power Ramp for Accelerated PreDistortion Training

Abstract

The high Peak to Average Ratio (PAR) of current and future wireless modulation standards imposes severe trade offs in the form of linearity versus efficiency for the transmitter.  Adaptive PreDistortion (APD) is often adopted as a solution to this trade off problem.  However, standards such as 3GPP, do not explicitly facilitate training of the APD Look Up Table (LUT) which can compromise its performance from a ‘blind start’ and following changes in the transmitter nonlinear characteristics due to, for example, a transient load change.  The following publication details how the relatively less prescriptive section of the 3GPP standard can be exploited to insert dedicated training waveforms to accelerate the training of the APD LUT.  In particular, the inter-slot boundary, power transient AM ramping waveform can be modified to improve training near the lower and upper enters of the LUT which otherwise would receive poorer training if the natural modulation waveform was solely rely on.

Body

The evolution of wireless standards towards spectrally more efficient modulation schemes has resulted in increasingly higher Peak to Average Ratio (PAR) in the transmit AM waveform.  To maintain the necessary linearity in the presence of a higher PAR the conventional approach is to ‘back off’ the PA relative to its compression point.  By biasing the PA to run below its compression point, a limited or controlled level of distortion can be achieved thereby preserving the linearity of the transmitter.  However, the principal disadvantage of this approach is efficiency.  Higher PAR requires larger back off which in turn reduces PA efficiency, ultimately shortening the on time of the mobile device. 

Adaptive PreDistortion (APD) has been adopted as a solution to the problem of linearity versus efficiency in high PAR modulation schemes.  Critical to the successful application of APD is the rapid training of the Look Up Table (LUT) which stores the inverse nonlinear gain of the PA.  Standards such as 3GPP do not explicitly facilitate dedicated training of such LUTs and hence the APD’s performance may be compromised, particularly when starting from a ‘blind’ state or when the PA nonlinear characteristics change due, for example, to a load transient.  The problem therefore exists of rapidly training the APD LUT in the presence of such unknown and time-varying states while remaining compliant to the governing specification.  Figure 1.1 illustrates a typical case for an EDGE modulation scheme.  A blind start where no a priori knowledge of the predistortion gain is assumed and the LUT entries (12 off) are initialized to unity.  Then, over 4 consecutive slots the natural modulation or unmodified EDGE waveform is used in the training...