BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Engineering Mathematics and Programming

[Signals and Systems]

Module Leader: Dr Claire Lines

Lecturer: Dr Alexey Pichugin

Authors: Professor Paul Sermon

Dr Alexey Pichugin

Dr Matthias Maischak

with illustrations from S. Haykin & B. van Veen’s “Signals and Systems” [TK5102.5.H38]

Brunel University London Page 1

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

SIGNALS AND SYSTEMS

Generally, systems accept/manipulate/generate signals (analogue or digital) with different purposes. We need to develop a
language to talk about such systems. Our view of systems will be from an input-output perspective, where a system responds to
applied input signals (which can be either discrete or continuous), and its response is described in terms of one or several signals:

Take note of the notation:

• 𝑥(𝑡) and 𝑦(𝑡) denote continuous-valued variables (and CT stands for “continuous time” signal or process);
• 𝑥[𝑡] and 𝑦[𝑡] denote discrete-valued variables (and DT stands for “discrete time” signal or process);
• We would call 𝑥(𝑡), or 𝑥[𝑡], independent variables, and 𝑦(𝑡), or 𝑦[𝑡], dependent.

Brunel University London Page 2

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

The separation between CT and DT can be somewhat arbitrary. Many signals we want to process are essentially continuous:

Brunel University London Page 3

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

It is common to discretize CT signals (this is usually called “sampling”) and then work with them using DT approaches:

Other signals can be essentially discrete (e.g., it is often the case in economics and finance), but CT could be a better setting for
modelling some of these phenomena.

There are many other examples of systems with interesting signals:

• Stock prices vary all the time - that’s what day-traders are trying to exploit and develop algorithms to analyse financial and
economic factors to predict stock prices;
• Cruise control in a modern car operating via throttle feedback;
• Chemical reactor that oscillates in rates with time;
• Stability control via a MEMS (microelectrical mechanical system) accelerometer;
• Dynamics of an aircraft or space vehicle;
• An edge detection algorithm for medical images.

Brunel University London Page 4

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

MORE COMPLEX SYSTEMS

All examples thus far were SISO (single-input single-output), but MIMO (multi-input multi-output) systems are also important

Did you notice that we only considered what system components do, not how? Effectively, we treat the components as “black
boxes”, to indicate that we do not care much for their internal specifics. We only consider functionality of the components, and
then use their interconnections to direct the flow of information through systems.

Therefore, another important concept is that of

interconnecting systems:

• We can build more complex systems by

interconnecting simpler subsystems;
• We can also use interconnections to modify the
response of an existing system.

The following signal flow (block) diagrams show three key types of
interconnections that can be used:

Brunel University London Page 5

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

(a) Cascade interconnection

The cascade interconnection is a successive application of
The cascade interconnection is a successive application of two (or more) systems on an input signal:
two (or more) systems on an input signal:

Mathematically, we can say that 𝑦 = 𝐺2 (𝑦1 ) or 𝑦(𝑥) = 𝐺2 (𝐺1 (𝑥)).

(b) Parallel interconnection

The parallel interconnection is an application of two (or more) systems to the same input signal, and the output is taken as the
sum of the outputs of the individual systems

In this case we can say that 𝑦 = 𝑦1 + 𝑦2 or 𝑦(𝑥) = 𝐺1 (𝑥) + 𝐺2 (𝑥). It is sometimes useful to write the second equation
differently, to highlight the parallel nature of the processing, as 𝑦(𝑥) = (𝐺1 + 𝐺2 )(𝑥).

Brunel University London Page 6

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

In both cascade and parallel interconnections, the signal flows through each one of them in the forward direction only. This
does not have to be the case in every situation, and many practical systems make use of feedback interconnections.

(c) Feedback interconnection

The feedback interconnection of two systems is a feedback of the output of system 𝐺1 to its input, through system 𝐺2 . In this
context, an error signal 𝑒 characterizes the error between a desired output signal and a direct measurement of the output.

Mathematically, 𝑦 = 𝐺1 (𝑒) and 𝑒 = 𝑥 − 𝐺2 (𝑦).

This is an example of a practical system with feedback that all of you would probably recognize:

Brunel University London Page 7

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Of course, a real system contains combinations of several (possibly many) interconnections, of different types, e.g.

At the top level we have a DT system 𝑦[𝑛] = 𝐺(𝑥[𝑛]).

However, internally, we have the cascade interconnection of the system that generates 𝑠[𝑛] and 𝐺4 , indeed 𝑦[𝑛] = 𝐺4 (𝑠[𝑛]).
Here 𝑠[𝑛] is produced by the parallel interconnection of the system generating 𝑤[𝑛] and 𝐺3 , so 𝑠[𝑛] = 𝑤[𝑛] − 𝑧[𝑛], with
𝑧[𝑛] = 𝐺3 (𝑥[𝑛]), and 𝑤[𝑛] is produced by the cascade interconnection of 𝐺1 and 𝐺2 .

Overall then,

𝑦[𝑛] = 𝐺4 (𝑠[𝑛]) = 𝐺4 (𝑤[𝑛] − 𝑧[𝑛]) = 𝐺4 (𝐺2 (𝑣[𝑛]) − 𝐺3 (𝑥[𝑛])) = 𝐺4 (𝐺2 (𝐺1 (𝑥[𝑛])) − 𝐺3 (𝑥[𝑛])).

Brunel University London Page 8

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

MATHEMATICAL OPERATIONS ON SIGNALS

Operations performed on dependent variables

Suppose that 𝑥(𝑡) is a continuous-time signal. We say that signal 𝑦(𝑡) is obtained by amplitude scaling of 𝑥(𝑡) if

𝑦(𝑡) = 𝑐𝑥(𝑡),

and call constant 𝑐 a scaling factor. The same operation can be applied to discrete-time signals:

𝑦[𝑛] = 𝑐𝑥[𝑛].

Given two CT signals 𝑥1 (𝑡) and 𝑥2 (𝑡), we can consider the signal obtained by addition:

𝑦(𝑡) = 𝑥1 (𝑡) + 𝑥2 (𝑡).

For example, when music is mixed on a mixing deck, signals from individual channels are added to each other.
Have you ever listened to AM radio? AM radio signal is obtained by multiplication:

𝑦(𝑡) = 𝑥1 (𝑡)𝑥2 (𝑡).

Of course, addition and multiplication can be applied to both CT and DT signals. To give you an example of an operation that
can only be applied to CT signals: 𝑦(𝑡) can be obtained by the differentiation of signal 𝑥(𝑡) with respect to time:
𝑑
𝑦(𝑡) = 𝑥(𝑡).
𝑑𝑡

Brunel University London Page 9

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Operations performed on independent variables

One important operation is called time-scaling. Given a CT signal 𝑥(𝑡), we say that signal 𝑦(𝑡) is obtained by time-scaling if

𝑦(𝑡) = 𝑥(𝑎𝑡).

• When 𝑎 > 1, the signal 𝑦(𝑡) is a compressed version of 𝑥(𝑡);

• When 0 < 𝑎 < 1, the signal 𝑦(𝑡) is an expanded (or stretched) version of 𝑥(𝑡).

For example, we can use time-scaling to narrow or broaden a peak:

For DT systems only integer values of 𝑎 make sense, so you can only compress, and, when compressing a DT signal, values of
𝑎 > 1 result in samples being lost.

Brunel University London Page 10

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Instead of scaling the time, we can consider its reflection, i.e. the replacement of 𝑡 by – 𝑡:

𝑦(𝑡) = 𝑥(−𝑡).

Time-shifting is also possible, for which

𝑦(𝑡) = 𝑥(𝑡 − 𝑡0 ).

Brunel University London Page 11

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

PROPERTIES OF SYSTEMS
Causality: A system is said to be causal if present output signal(s) depend only on the present or past input signal(s).

Example 1:

Is system 𝑦(𝑡) = 𝑥(𝑡 − 1) causal or non-causal?

Answer: consider several values of time 𝑦(1) = 𝑥(0), 𝑦(2) = 𝑥(1), etc. You can see that 𝑦(𝑡) always depends on past values
of 𝑥(𝑡), hence this system is causal.

Example 2:

Is system 𝑦(𝑡) = 𝑥(𝑡 + 1) causal or non-causal?

Answer: again, consider several values of time 𝑦(1) = 𝑥(2), 𝑦(2) = 𝑥(3), etc. You can see that 𝑦(𝑡) always depends on future
values of 𝑥(𝑡), hence this system is non-causal.

Memory: A system is said to possess memory if the output signal depends on past or future input signals. A system is
memoryless if its output at time 𝑡 depends only on the input at the same time.

Hopefully you can see that both systems in Examples 1 and 2 possess memory.

Brunel University London Page 12

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Stability: A system is said to be bounded-input bounded-output (BIBO) stable if for any bounded input, the corresponding
output is also bounded.

Invertibility: A system is said to be invertible if the input signal can be recovered from the output signal.

In an invertible system you can determine its input signal 𝑥(𝑡) uniquely by observing its output signal 𝑦(𝑡).

Example:
1 1
Consider cascade interconnection where 𝑦1 (𝑡) = 2𝑥(𝑡). If we choose 𝑦(𝑦1 ) = 𝑦1 , then 𝑦(𝑡) = 2𝑥(𝑡) = 𝑥(𝑡).
2 2

Clearly, 𝑦1 (𝑡) is an example of an invertible system.

Brunel University London Page 13

 BE1608 ENGINEERING MATHEMATICS AND PROGRAMMING

Time-invariance: A system is called time-invariant (TI) if its behaviour does not depend on what time it is.

Mathematically, for DT systems, we can say that system 𝑥[𝑛] → 𝑦[𝑛] is time-invariant if for any input 𝑥[𝑛] and any time shift
𝑛0 it is true that

Similarly, for a CT time-invariant system it is true that

The systems that are not time-invariant are called time-varying.

Example:

Consider the following system: 𝑦(𝑡) = sin(𝑥(𝑡)). Is it time-invariant or time-varying?

Answer: Let us denote by 𝑦1 (𝑡) the output produced by the shifted input 𝑥1 (𝑡) = 𝑥(𝑡 − 𝑡0 ).

We have 𝑦1 (𝑡) = sin(𝑥1 (𝑡)) = sin(𝑥(𝑡 − 𝑡0 )). At the same time, 𝑦1 (𝑡 − 𝑡0 ) = sin(𝑥(𝑡 − 𝑡0 )) = 𝑦1 (𝑡).

Hence, the system is time-invariant.

Brunel University London Page 14