Rate adaptation using error vector magnitude

• US 7,826,521 B1
• Filed: 02/28/2007
• Issued: 11/02/2010
• Est. Priority Date: 02/28/2007
• Status: Expired due to Fees

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Atheros Communications Incorporated (Qualcomm, Inc.)
• Inventors: Petrus, Paul, Sun, Qinfang
• Primary Examiner(s): Fan; Chieh M
• Assistant Examiner(s): Garcia; Santiago
• Application Number:
  US11/680,535
• Time in Patent Office: 1,343 Days
• Field of Search: 375/225
• US Class Current: 375/225
• CPC Class Codes: H04L 1/0003 by switching between different modulation schemes
• CPC Class Codes: H04L 1/0019 in which mode-switching is based on a statistical approach

Abstract

Rate adaptation is accurately provided by using an average SNR technique, a predicted PER technique, or an EVM to PER mapping technique. An average SNR is computed by averaging EVM values in a particular domain. Using the computed average SNR, an optimized rate is determined. The predicted PER technique includes computing EVM values for a current packet, determining an average BER using these EVM values for each data rate, determining an average SNR from this average BER, using the average SNR and a PER versus SNR curve to determine the predicted PER for each data rate, and using a data rate and the predicted PER to compute the predicted throughput for that data rate. In the last technique, an EVM is mapped to a packet error rate (PER) for each data rate. A throughput is computed for each data rate and PER. A data rate with a high throughput is selected.

Claims

1. A method of adapting a rate for a radio frequency device, the method comprising:

in a radio frequency device computing an average signal-to-noise ratio (SNR) from a plurality of error vector magnitude (EVM) values; and

using the average SNR and a (SNR, rate) lookup table to determine an optimized rate,

wherein computing the average SNR includes;

computing a bit error rate (BER) for each EVM value;

computing an average BER for all EVM values; and

determining an SNR corresponding to the average BER,

wherein determining the average BER includes computing a local average BER for a de-interleaved bit sequence using a sliding window format.

2. The method of claim 1, wherein the EVM values include at least one of pilot EVM values and data EVM values.

3. A method of adapting a rate for a radio frequency device, the method comprising:

in a radio frequency device computing error vector magnitude (EVM) values for a current packet;

determining an average bit error rate (BER) using the EVM values for each data rate;

determining an average signal-to-noise ratio (SNR) from the average BER;

for each data rate, determining a predicted packet error rate (PER) using the average SNR and a PER versus SNR curve;

for each data rate, computing a predicted throughput using the data rate and the predicted PER; and

selecting a rate having a high predicted throughput,

wherein determining the average BER includes computing a local average BER for a de-interleaved bit sequence using a sliding window format.

4. The method of claim 3, further including generating PER versus SNR curves, which comprises:

simulating PER versus SNR curves for an additive white Gaussian noise (AWGN) channel for all data rates; and

saving the PER versus SNR curves.

5. The method of claim 3, wherein the PER versus SNR curve is generated from an additive white Gaussian noise (AWGN) channel.

6. The method of claim 5, wherein determining the average BER includes averaging a BER over all bins with EVM values, thereby providing a global average BER.

7. The method of claim 6, wherein the EVM values include at least one of pilot EVM values and data EVM values.

8. The method of claim 5, wherein computing the local average BER comprises:

for each data bin, computing a BER of each data bit using data EVM values, a modulation type, and a bit position;

de-interleaving all BER values; and

computing the local average BER in a the sliding window format using de-interleaved BER values.

9. A method of adapting a rate for a radio frequency device, the method comprising:

in a radio frequency device for each rate, mapping an error vector magnitude (EVM) to a packet error rate (PER);

computing a throughput for each data rate and PER; and

selecting a data rate with a high throughput,

wherein mapping the EVM to the PER includes;

estimating EVMs of pilots for each data symbol in a packet;

computing an average EVM per pilot across all data symbols in the packet;

estimating a mean EVM and a standard deviation for each stream;

computing the mean EVM and the standard deviation for all streams;

computing an effective EVM; and

using a lookup table to map the EVM to the PER.