# Does Maximum Likelihood detector (QAM symbol detection) enhance BER?

Assuming QAM symbols (or OFDM system) with AGWN, I found that the use of ML detector does not enhance the BER but only enhances (reduces) error vector magnitude (EVM) because of symbol decision and matching each received (corrupted) QAM constellation point to the nearest ideal QAM constellation point.

I am a bit confused as I expected the ML detector to enhance both BER and EVM not just EVM? Thanks

• "Enhances" compared to what? If something else is as good as the ML detector, congratulations, you have found an ML detector.
– mmmm
Jun 12 at 9:50
• compared with the theoretical BER calculation, or simply compared with the case where ML detector is not applied. Note: I am using matlab command (biterr) for BER simulation result Jun 12 at 9:53
• What do you mean with "theoretical BER calculation"? What decider do you assume while calculating? Or are you referring to Shannon capacity?
– mmmm
Jun 12 at 10:17
• I will post matlab code to help to understand my question Jun 12 at 10:33
• Assuming the transmitted symbols are independent and equally likely, the matched filter followed by a threshold detector is ML and therefore it's also optimum.
– MBaz
Jun 12 at 14:39