Browse Prior Art Database

SNR BASED ANALOG DIVERSITY SYSTEM

IP.com Disclosure Number: IPCOM000007183D
Original Publication Date: 1994-Jun-01
Included in the Prior Art Database: 2002-Mar-04
Document File: 3 page(s) / 137K

Publishing Venue

Motorola

Related People

Brad Hiben: AUTHOR [+3]

Abstract

Systems employing analog diversity voting con- sist of an analog comparator and more than one receiver. Each receiver demodulates the analog audio and sends the baseband audio down a dedicated phone line to the comparator. The comparator should then select the phone line which has the signal with most intelligibility.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 3

MO-LA Technical Developments Volume22 June 1994

@

SNR BASED ANALOG DIVERSITY SYSTEM

by Brad Hiben, Bob LoGalbo and Don Newberg

1. INTRODUCTION: phone line to the comparator. The comparator should then select the phone line which has the signal with Systems employing analog diversity voting con- most intelligibility.
sist of an analog comparator and more than one
receiver. Each receiver demodulates the analog audio A typical system configuration employing a cen- and sends the baseband audio down a dedicated tralized comparator is depicted below:

Comparator

Figure 1: Typical Comparator Based System

improvement lies.

  The existing method of analog voting high pass filters the baseband signal to attenuate most of the speech energy, The signal is then half-wave recti- fied, and passed through a peak detector and a voice peak rejecter. This method very accurately selects the signal with the better SNR when voting among signals with high SNR's. When a signal with a low SNR is being voted upon, it is most likely to be voted even in the presence of signal's with higher SNR's because ofreasons specific to the algorithm.

When signals with low SNR's are being voted

upon, it is crucial to vote the signal with the better SNR. If an analog signal has an average SINAD of 25dB or higher, it is generally intelligible. Therefore it isn't crucial to select a signal with a 30dB SINAD signal over a 26dB SINAD signal when concerned only with intelligibility. But it is crucial to select the signal with the 19dB SINAD over the signal with 15dB SINAD when concerned with intelligibility. This region of SNR's is where a need for voting

0 Motorola, 1°C. 1994

2. SOLUTION:

  The solution is an alternative method for voting on analog baseband signals. The new voting algo- rithm consists of autocorrelation and crosscorrelation calculations and some algebra. The hardware con- figuration stays the same, but the voting algorithm within the comparator is changed.

Assume the following three inputs are present at the comparator:

1) s,=d,+n,
2) s, = Ad, + n2
3) s, = Bd, + n3

where: s,, s2, & s, are discrete sequences sampled from ports 1, 2 & 3. d, is the inbound subscriber signal (discrete).

41

[This page contains 15 pictures or other non-text objects]

Page 2 of 3

0 M MO-LA

Technical Developments Volume 22 June 1994

A,B are gain values unique to the individ- ual phone lines.
n,, n2, and n3 are independent discrete zm. noise variables.

11) r, = E[s& = A'E[d,q + E[n,q

Therefore:

12) SNR,= ' (r,k,-l)A,

where A = k,c,,

  Note that it is assumed that the noise variables are uncorrelated from receiver to receiver and that the differential delay between signals arriving at the comparator is much less than the sampling rate. The signal q is the same voiced audio in all 3 cases except for the gain terms and added noise.

With these definitions, the following crosscorrelations would be true:

Continuing to find SNR, for the th...