Creation of Real World Test Signals for ATSC Mobile DTV Receiver Performance Examined at NAB Show

Each year, presenters at the NAB Broadcast Engineering Conference are urged to submit written papers on the subject of their presentations. The yearly NAB Broadcast Engineering Conference Proceedings, published as both a book and a CD-ROM, and sold at the NAB Store in the Las Vegas Convention Center during the NAB Show - and at www.NABStore.com after the Show - is a compendium of these technical papers, and an important archive of the leading edge of broadcast engineering issues. The NAB Best Paper Award, established in 2010, honors the author(s) of a paper of exceptional merit published in the Proceedings, and is announced at the annual Technology Luncheon. At this year's Show, the award was given for the paper "Creation of a Library of RF Recorded Signals for ATSC Mobile DTV in a Single Frequency Network Environment" authored by Charles Nadeau, Benoit Ledoux, Yiyan Wu, Gilles Gagnon, Robert Gagnon and Sébastien Laflèche, of the Communications Research Centre, Canada. The paper is excerpted below.

INTRODUCTION
With the growing interest in Single Frequency Networks (SFN) as a possible solution to increase reception coverage for mobile receivers and handheld devices, it is desirable to have a reliable and repeatable way of testing ATSC-M/H receivers' performance in mobile SFN environments. One such way is to test the ATSC-M/H receivers by using a library of well characterised RF captured field ensembles. The Communications Research Centre Canada (CRC) took the initiative of working on such a library, which contains RF captures of SFN mobile channels from areas with various types of population density, vehicle speed and carrier-to-noise ratio. This article describes how the RF captured field ensembles were recorded, selected and evaluated. Furthermore, this is followed by an analysis demonstrating how using real field captures to test current generation ATSC-M/H receivers yield interesting results that were not apparent during standard laboratory characterisation of these receivers.

ATSC-M/H SFN TRANSMISSION SYSTEM
The transmission system used in the creation of the RF captured field ensembles is a Single Frequency Network located in Ottawa, Ontario, and operated by CRC for television broadcast research. Four low-power transmitters cover the Ottawa region and transmit the ATSC-M/H signal on UHF channel 46. Figure 1 illustrates the topology and power output of the four transmitters, with lighter areas indicating reception power over -80 dBm, as predicted by CRC-COVLAB [1].

Recording the RF signal during the campaign simply involved driving along the chosen routes in each region of interest, respecting speed limits and road signalization. Depending on the area, the routes took between 2 to 25 minutes. Only the transmitters that were identified to produce challenging SFN reception conditions for each test route were turned on during a drive, as explained in the above section. The transmitters' outputs were controlled remotely from the test vehicle. A monitoring test site was also used to verify the transmitters' output power.

III. RF Recorder Hardware Considerations and Setup
Capturing RF signals correctly involves using a test setup that introduces he least amount of distortions to the captured signal while preserving important signal characteristics such as fast power variations and carrier-to-noise ratio (C/N). RF recorder hardware has its limitations in that the tuner has a noise figure and the digitizer has a limited dynamic range. A good recording system usually uses a slow automatic gain control (AGC) to bring the signal within optimal range of the digitizer, as to not overload the system and to maximize C/N. However, this brings its own problems such as the RF recorded power not being the real signal power when there are fast power variations (like in a mobile environment) because the AGC loop response time can impact the detected RF power. Moreover, using an AGC also affects the RF recordings' noise floor for low-power signals since the front end' noise floor becomes dominant and can vary when the AGC chain of amplifiers and attenuators changes states to respond to fast RF power variations that occur during mobile reception.

With this in mind, it was deemed important for this library of RF recordings to preserve signal power variations and to have a known noise floor that would become the noise floor that eventual receivers would see when the RF sequence would be played back. The RF recorder was thus setup with a fixed front end gain so the operating dynamic range at the digitizer is known and fixed. The system was then characterized to know this noise floor and to know the maximum input RF power without overloading the front end or the digitizer. This is a summary of the RF recorder setup:

o Sampling rate: 12.5 MS/s, IQ samples
o 10 MHz of effective bandwidth around the center frequency of 665 MHz (channel 46).
o No automatic gain control
o RF Recording Noise Floor: -94 dBm
o Maximum input power: -45 dBm

SELECTION OF RF CAPTURED FIELD ENSEMBLES

Following the campaign, six hours of raw RF recordings was available to build the library of RF captures. A pre-analysis was done in order to identify interesting portions within the raw data. It was decided that the RF captures should be 2 minutes in duration to be able to thoroughly analyse receivers' behavior over a mobile channel. The analysis was based on a combination of the following three evaluation characteristics considered over the 2 minutes duration:

o Carrier-to-Noise ratio
o Multipath channel profile (impulse response)
o Reception performance of the RF recordings when the signal is played back to a reference receiver.

After this pre-analysis, approximately 150 candidate RF captures were identified. From this pool of captures, the final selection was made to have a manageable number of captures that would cover the range of reception difficulty for all three services that were carried over the ATSC M/H signal. Thus, there would be captures that should be difficult for a receiver only for the ½ rate coded service, but easy for the more robust modes, and so on. Table 1 describes the difficulty categories that were used to classify the RF captures when testing them on an ATSC-M/H receiver.

*A single small burst less than 5 seconds during the RF playback is considered "No errors"

The final selection for the library of ATSC-M/H RF recordings includes 45 RF captures of 2 minutes length each evenly spread across the difficulty categories described in Table 1, considering reception performance for the best mobile DTV receiver available. The 45 chosen RF captures include many types of reception environments such as urban and suburban building density and different types of road speeds such as the stop-and-go drive of downtown areas, moderate speeds with occasional stops of suburban areas and high speed freeway drive. Table 2 shows how many RF captures fall in each general type of reception environment.

The complete paper is included in its entirety in the 2012 NAB Broadcast Engineering Conference Proceedings, available online from the NAB Store (www.nabstore.com). Other technical papers related to mobile DTV published in the Proceedings include the following: