I am studying data communication and networking. While reading about data transfer using electrical signals, I came across 'frequency division multiplexing'. I understood how data is transferred in the form of electrical signals i.e. by way of modulating frequency, amplitude etc. Understanding a one way transmission is easy but I am having difficulty grasping the concept of two-way simultaneous transmission as well as multiple transmissions on high speed lines :( I am wondering how multiple electrical signals are transferred without disturbing each other. Don't they interfere into each others frequency ?
It's best to view this in the frequency domain rather than the time domain.
Here is a basic example - two signals SIG (~1 MHz) and SIG2 (~2 MHz) are combined to form SIG+SIG2, then passed through LC filters with corresponding resonant frequencies to separate them again:
Signals in and combined:
Signals Out after filters:
FFT of SIG + SIG2, and FFT after each filter:
Note the difference between SIG+SIG2 in the frequency domain as opposed to the time domain (blue waveform in second plot) - the two separate frequencies are much easier to see.
There are a few ways that engineers have devised to have multiple "channels" on the same medium. Consider circuit switching vs. packet switching. In a circuit switched telecom network a dedicated communications channel between two devices is established. All data is continuously transmitted while the channel is active (Visualize this as "analogue"). In a packet switched network, messages are divided into discrete packets which are then transmitted (Visualize this as digital, like computer networks)
When considering electric circuits we have two domains in which we work, namely the frequency- and time-domain. I am not going to go into the mathematics at work here but the Laplace transform is used to transform time signals into frequency signals and vice versa.
From these two domains there can be derived two methods for simultaneous communication. Firstly you can keep the frequency band used for your signal constant, and divide your time signal into slots, during which you transmit parts of different messages. This is known as time division multiplexing (TDM). On the other hand you can use different frequency slots and keep the time signal out of the equation, known as frequency division multiplexing (FDM).
Lost me? This is how TDM works:
Time is divided into slots, each slot is used to transmit a data block of each channel that is being used. Block 1 is usually a synchronization block.
FDM: Consider two message signals with arbitrary frequency spectra 1 and 2. By assigning each signal with a separate band to operate in, we can send the two signals simultaneously and later retrieve them by using frequency selective filters.
So in order to answer your question, no, we can make it so that different signals occupy different bands of the frequency spectrum. It is however necessary that these signals be band limited (Finite spectrum width).
There are quite a lot of other methods to achieve simultaneous transmission, including polarization, MIMO, and then a whole bunch of packet switched methods (Although you can argue that packed switched is not simultaneous).
As long as the transmission medium is linear, ie 1v + 1v = 2v, you can have multiple transmitters sending different patterns which will linearly add in the medium, but can be sorted back out again by a receive "filter" which looks for the unique pattern of the corresponding transmitter.
In conventional radio and comparable frequency division multiplexing over wires, the pattern used is a sinusoid of a particular frequency. Simplistically speaking, the transmitter sends it or doesn't (there are of course more complicated encodings) and the receiver sees the presence of that frequency in the medium or doesn't.
Modern electronics - especially digital techniques - mean that the unique pattern isn't constrained to be a sinusoid, but can be more complicated sequences. Instead of a frequency, it can be a "secret sequence" - either literally secret, or just obscure enough to be different from what everyone else is using.
But if the medium is non-linear, such that 1v + 1v = 1.9v, then information transmitted with different frequencies or patterns does not remain independent and able to be cleanly filtered back out, instead mutual interference results. This doesn't tend to happen with radio waves themselves, but it tends to happen to greater or lesser degree in almost all electronic components - the more one (strong) signal pushes the instantaneous total of the combination of signals into the territory where the component can no longer operate in a quite linear fashion, the more it and other signals will be slightly shortchanged and produce distortion at other frequencies. A particularly bad case is when an amplifier is over-driven to the point where the sine waves start getting abruptly clipped to flat tops - this produces interference at numerous frequencies - but the problem begins when the degree of flattening is still too subtle to even be seen on an oscilloscope. Another common interference-generating non-linear effect is rectification, if the signal is passed through a diode, either an intentional one or an un-intentional one formed from a metal oxide coating on a mechanical connection between conductors.
Not quite the same but close enough ...
Hold two classical "tuning forks" in one hand with metal handles tightly touching.
Tap the tines of one fork and it will "ring".
Tap the other and it too will ring.
The two are tightly joined mechanically but are isolated for frequencies that they are designed to oscillate at.
Clamp all these forks to a solid steel bar and they will still ring correctly.
It's analogous to how broadcast radio works - just the medium is different. You can transmit and receive simultaneously on multiple different channels so long as the bandwidth of each channel is suitably constrained.