different researchers associate sleepiness (or sleep onset) with parameters from frequency domain, e.g.:
Article Prediction of fatigue and sleep onset using HRV analysis
Article Heart Rate Variability Can Be Used to Estimate Sleepiness-re...
Article Autonomic changes during wake-sleep transition: A heart rate...
However, the findings concerning HRV are not always consistent and a prediction might be difficult (amongst other things owing to the many factors affecting HRV).
Heart Rate Variability can be evaluated by the time domain(SDNN, SDANN,RMSSD, NN50,pNN50 etc)and frequency domain ( LF, HF VLF )measures.The values of SDNN and RMSSD increases during the NREM sleep mainly during Deep sleep (N3) when compared to the wakefulness and REM sleep states. These values are usually higher in athletes, people doing guided exercises like deep breathing techniques. The values of the SDNN and RMSSD varies from person to person.
In frequency domain, Lower frequency (LF) is used to show the influence of combination of sympathetic and parasympathetic activities whereas High Frequency (HF) only infers the parasympathetic activity. To infer the sympathetic activity researchers usually use the ratio of LF/HF so that parasympathetic power cancels out of the ratio. In NREM sleep parasympathetic activity is high as measured by HF(HF increases) whereas sympathetic activity is high during REM and wake state ( decrease in HF) sleep as measured by ratio of LF/ HF.
Athletes, people doing guided exercises like deep breathing are mostly parasympathetic dominant even during wakefulness and REM sleep.
You can increase your parasympathetic activity just before going to bed to fall asleep as soon as possible by doing deep breathing exercises, passive heating (take warm bathing), drinking water