It turns out that if we know what kind of signal to expect, we can better receive those signals. This should be intuitive, because it is hard to find something if we don't know what precisely we are looking for. How is a receiver supposed to know what is data and what is noise, if it doesnt know what data looks like?
Coherent transmissions are transmissions where the receiver knows what type of data is being sent. Coherency implies a strict timing mechanism, because even a data signal may look like noise if you look at the wrong part of it. In contrast, noncoherent receivers don't know exactly what they are looking for, and therefore noncoherent communication systems need to be far more complex (both in terms of hardware and mathematical models) to operate properly.
This section will talk about coherent receivers, first discussing the "Simple Receiver" case, and then going into theory about what the optimal case is. Once we know mathematically what an optimal receiver should be, we then discuss two actual implementations of the optimal receiver.
It should be noted that the remainder of this book will discuss optimal receivers. After all, why would a communication's engineer use anything that is less than the best?
The Simple ReceiverEdit
A simple receiver is just that: simple. A general simple receiver will consist of a low-pass filter (to remove excess high-frequency noise), and then a sampler, that will select values at certain points in the wave, and interpolate those values to form a smooth output curve. In place of a sampler (for purely analog systems), a general envelope filter can also be used, especially in AM systems. In other systems, different tricks can be used to demodulate an input signal, and acquire the data. However simple receivers, while cheap, are not the best choice for a receiver. Occcasionally they are employed because of their price, but where performance is an issue, a better alternative receiver should be used.
The Optimal ReceiverEdit
Mathematically, Engineers were able to predict the structure of the optimal receiver. Read that sentence again: Engineers are able to design, analyze, and build the best possible receiver, for any given signal. This is an important development for several reasons. First, it means that no more research should go into finding a better receiver. The best receiver has already been found, after all. Second, it means any communications system will not be hampered (much) by the receiver.
here we will attempt to show how the coherent receiver is derived.
The matched receiver is the logical conclusion of the optimal receiver calculation. The matched receiver convolves the signal with itself, and then tests the output. Here is a diagram:
s(t)----->(Convolve with r(t))----->
This looks simple enough, except that convolution modules are often expensive. An alternative to this approach is to use a correlation receiver.
The correlation receiver is similar to the matched receiver, instead with a simple switch: The multiplication happens first, and the integration happens second.
Here is a general diagram:
r(t) | v s(t) ----->(X)----->(Integrator)--->
In a digital system, the integrator would then be followed by a threshold detector, while in an analog receiver, it might be followed by another detector, like an envelope detector.
To do the best job of receiving a signal, we need to know the form of the signal that we are sending. This should seem obvious, we can't design a receiver until after we've decided how the signal will be sent. This method poses some problems however, in that the receiver must be able to line up the received signal with the given reference signal to work the magic: If the received signal and the reference signal are out of sync with each other, either as a function of an error in phase or an error in frequency, then the optimal receiver will not work.