Simon is right except there is one more assumption required. If the transmitter and receiver are moving toward or away from each other, then Doppler effects will cause the pulses to be closer together or farther apart in time.
One other kind of weird thing to consider. The channel between TX and RX could be time varying. The propagation speed of EM waves depends on the dielectric properties of the space in between. If a large dielectric solid material was somehow inserted between TX and RX after the first pulse, then the second pulse would be substantially delayed due to the slower EM propagation in the dielectric material. I don't think it is possible to do this in the space of 5ns, but since I don't know what your final application is, I thought I would mention it. Other things like humidity or rain or snow can have an effect also, but normally these are not high-speed processes.
I suspect multi-path will be your biggest problem. You will need to look into some kind of correlation detection scheme.
Also, you will have a hard time getting approval from FCC or similar agencies. In other words, have fun experimenting, but don't think you are going to market a product in the developed world based on broadband pulse transmission unless you are willing to go through a lot of trouble.
One last thing. The shape of your received pulse will depend somewhat on the channel bandwidth, which will depend on antenna bandwidth. The more bandwidth your antennas have, the closer the RX pulse will look to the TX pulse.
There are two roads: one that lets you understand what's happening, and another that is fast. Let's start with the firts.
please note: I'm assuming 1Mbitps = \$2^{20}\$bps and not \$10^6\$bps.
Your transmitter would divide the bitrate across 32 modulators: each modulator would see a reduced bitrate, i.e. 1Mbps/32 = 32768bps = 32kbps. Since you have two amplitude levels the baud rate and the bit rate correspond: each of your modulators works with a 32kbaud rate needing 65.536kHz of bandwidth (of noiseless channel, Shannon would add). Your total bandwidth is then your number of channels times each channel's bandwidth, i.e. 64kHz*32 = 2.1MHz (approx.).
Now for the fast way: you have a bit rate of 1Mbps, all your modulator's symbols are the same, i.e. 1 bit per symbol, so you can just double your bit rate and get the bandwidth:
2*1Mbps = 2*1.048M = 2.1MHz
Best Answer
The word 'width' in Pulse Width Modulation is a bit misleading. PWM actually encodes an analog value as the ratio of High or On time ('pulse width') to the total period of a rectangular waveform.
VPW is another name for Pulse Length or Pulse Duration Modulation (PDM). It encodes an analog value in the absolute width of the pulse. The period between pulses is unimportant.
PDM is used in RC servos, which typically respond to a pulse width varying from 1 to 2ms. The nominal repetition frequency is 50Hz (20ms between pulses) but most servos will work between 40~70Hz - and some go up to almost 500Hz (the theoretical maximum repetition frequency of a 2ms pulse).