Receiver codec super set constellation generator
View Patent ↗A technique is proposed to accurately estimate the Network CODEC levels for each PCM code a server modem generates. These levels are affected by the digital impairments such as Digital attenuation PAD in the trunk, the Robbed Bit Signaling, the type of CODEC (μ-law or a-law—or non standard), and by analog impairments such as loop distortion, noise, inter-modulation distortion, echo. At client modem equalizer output good estimates for these levels are derived. By detecting RBS pattern of the trunk, and using averages of decode levels of similar RBS slots, more accurate data points are obtained. By further replacing these levels with the closest CODEC receive levels, good accuracy is obtained. Non-monotonic points are detected and eliminated. An upper limit is set for constellation points to avoid saturation of the receiver. IMD correction is applied to the decode levels. Ideal points that are not signaled, are added if possible. When PAD-detection or Codec detection fails, PAD is set to 0 dB and the constellation is based on originally received and averaged data points. Techniques are presented for V.90 type modem constellation generation.
1. A method comprising:
receiving the linear output of a network CODEC, the linear output being converted from coded data transmitted by a network server modem, the linear data comprising a plurality of data points in a predetermined number of slots;
averaging the linear data value for each digital code for each of the time slots to generate estimated real non-linear constellation points;
converting the estimated real non-linear constellation points using a first converting algorithm to map the estimated real non-linear constellation points to network CODEC linear output levels;
matching each mapped CODEC linear output level to the closest of a plurality of ideal CODEC output levels for a selected type of network CODEC; and
mapping the closest ideal CODEC output levels back to real non-linear constellation points.
2. The method of claim 1 , wherein the linear output of a network CODEC is received by an analog modem.
3. The method of claim 1 , wherein mapping the closest ideal CODEC output levels back to real non-linear constellation points uses an inverse of the first converting algorithm.
4. The method of claim 1 , wherein the linear output of the network CODEC is converted from PCM (pulse code modulation) data.
5. The method of claim 1 , wherein the pre-selected frame size comprises one of 6 slots, 12 slots, or 24 slots.
6. The method of claim 1 , wherein the first converting algorithm comprises:
detecting digital PAD attenuation; and
multiplying each linear data value by an estimated digital PAD attenuation to map the linear data values to CODEC output values.
7. The method of claim 6 , further comprising:
detecting inter-modulation distortion, and
if inter-modulation distortion is detected, applying an additional level dependent multiplication to the linear data values.
8. The method of claim 6 , wherein a failure in PAD detection is treated as a 0 dB PAD and raw averaged data is used as the real non-linear constellation points.
9. The method of claim 6 , wherein if CODEC detection fails, raw averaged data is used as the constellation points.
10. The method of claim 1 , wherein matching each mapped CODEC linear output level comprises:
detecting the type of the network CODEC; and
slicing the converted linear values to ideal CODEC output values.
11. The method of claim 1 , wherein averaging the linear data values further comprises:
grouping similar robbed bit signaling slots, and
averaging constellation points of the similar robbed bit signaling slots to reduce the number of real non-linear constellations.
12. The method of claim 11 , wherein averaging the linear data values further comprises:
averaging only non-robbed bit signaling slots.
13. The method of claim 1 , wherein linear data output of the network CODEC is according to output levels of one of:
a G711 A-law CODEC;
a 711 β-law CODEC ; or
a D4 channel bank CODEC.
14. The method of claim 1 , further comprising: limiting the largest constellation point to a level supported by hardware before saturation.
15. The method of claim 1 , further comprising: calculating and inserting ideal values that correspond to missing codes into the constellations when low level codes are not signaled due to statistical requirements and when line noise is small enough to support the low level codes.
16. The method of claim 1 , further comprising: eliminating constellation points that are non-monotonic.