Current control loop in PWM Inverter

I am working on a current controlled PWM inverter where I will use PI control to force the current through the inverter to follow a reference provided as is done in Active Power Filters, Grid connected inverters and in some forms of vector controlled ac drives.
I am currently stuck at getting a faithful measurement on the inverter current. I tried ACS712 yesterday but it's performance is just too low for such an application in terms of noise and speed.
I have ordered LEM sensors HX 10 and LTS 6-NP and hopefully they should do the trick.
Now could someone help me in terms of the following:
1) Let us say that I use an RL load R=4ohm and L=10mH which means Time constant L/R = 2.5ms (if I am not wrong). How do I decide PWM frequency?
2) I read that the sampling frequency for such current control loops is very high (about 30kHz) but I don't understand as wouldn't the PWM frequency be 12-16kHz max?
3) Am I right that sampling and calculations are done every PWM cycle? So where does this 30kHz come from?
4) So what should be my sampling frequency? Should it be the same as my PWM frequency?
5) What should be the cut-off frequency of the Anti Aliasing Filter and it's order and allowable phase lag in case of higher order filters?

I tried ACS712 yesterday but it's performance is just too low for such an application in terms of noise and speed.Can't understand this statement. I know a lot of PWM inverters that are performing closed loop current control with Allegro current sensors and similar products from other vendors.

Your question catalog seems to ask for retelling a power electronics text book. I doubt that the questions can be answered in a short post.

To start with the last question. You usually don't have anti-aliasing filters in a power electronics signal acquisition path. You primarly need to supress systematical errors caused by the pwm frequent current which can be achieved e.g. integrating/averaging over full pwm periods.

I will add my test circuit and waveforms for the ACS712. The issue is this, no matter what I do, the noise is never below say 250mV. Now my interest lies in the 0-3A range, for 3A, the output increases by 1.5V which means 500mV per Ampere which means that the noise of 250mV does not allow measurement below 0.5A.
Adding a low pass filter will be essential and when I just put a 0.1uF cap to see, noise dropped to less than 80mV which is of no use to me because the effective cut off frequency was well below the PWM frequency and my current waveform also got ruined.
So I need to know that for say 8kHz PWM, what can be the cut off frequency of the filter I add afer the sensor without which the noise is 0.5A or 1/6th the max value. Also this filter must not distort the actual signal.