%\documentstyle[fullpage]{article}
\documentstyle{article}
\oddsidemargin 0in
\evensidemargin 0in
\topmargin .25in
\textheight 8in
\textwidth 6.5in
 \parindent=0in
\parskip=0.5em
%\setlength{\textwidth}{6.5in}

\twocolumn
\pagestyle{myheadings}
\markright{\rm CS211 handout\hfill Embedded systems, distributed
computing, and threads\ \ \ }

%\newcommand{\emph}[1]{{\em #1}}
%\newcommand{\texttt}[1]{{\tt #1}}
\begin{document}


\section*{\bf Table of contents}
\ref{synopsis}. Introduction \dotfill \pageref{synopsis}\\
%
\ref{sec:microp}. Microprocessors\dotfill \pageref{sec:microp}\\
%
\ref{sec:realtimesim}. Real-time simulation\dotfill \pageref{sec:realtimesim}\\
%
\ref{sec:heating}. Simulating a heating system\dotfill \pageref{sec:heating}\\
%
\ref{sec:threads}. Threads in Java\dotfill \pageref{sec:threads}\\
%
\ref{sec:clock}. The SystemClock\dotfill \pageref{sec:clock}\\
%
\ref{sec:inside}. InsideTemperature\dotfill \pageref{sec:inside}\\
%


\section{Introduction}
\label{synopsis}

We (1) introduce you to the concept
of ``embedded systems'', which can consist of microprocessors that are
embedded in larger systems in order to control or sense physical
devices; (2) introduce you to ``distributed programming'', where
several different computers are executing
simultaneously, with some form of communication and synchronization
between them; and (3) tell you about {\em execution threads} in Java, 
which allow a Java program to consist of several program
segments executing in parallel.


The program that we demo can be obtained from the course web page 
(look under handouts).


\section{Microprocessors}
\label{sec:microp}

Microprocessors are small computers, usually embedded in other
systems. They are used to monitor and regulate the behavior
of many common commercial products.  E.g.~dedicated
microprocessor systems in cars regulate the fuel-air
mixture for combustion and prevent skidding (using anti-lock brake
systems).  In the home, they appear in appliances such as cellular
phones, microwave ovens and thermostats for heating and cooling
systems.  They are used in modems, laser printers, graphics boards,
and disk drives.  More exotic applications include medical
instrumentation, guidance of ``smart bombs'', detection of collision
and decay events in high energy particle accelerators.

Microprocessors appear in settings where {\em real-time computing\/}
is necessary: responses to external signals or changes ---which can
come at unpredicatable times--- must take place within a specific
amount of time.  Some microprocessors must also be able to generate
electrical signals that can be used to control other devices.

The anti-lock brake system on a car is a good example of the
challenging nature of real-time computing applications.  A
microprocessor must detect changes in the rotational rates of the
wheels characteristic of a skid.  If the angular velocity of the
wheels increases too fast, the tire is slipping on the pavement and
the car may be skidding; controlling the situation may be difficult
for the driver.  Detecting and correcting this situation involves
monitoring the rate of change of the angular velocity (i.e.~the
angular acceleration), not just the angular velocity itself. Upon
detecting this situation, the microprocessor must generate a sequence
of output pulses to control the hydraulic brake system,
allowing the wheels to intermittently rotate to prevent a skid, thus
making the car easier to control.  Clearly, quick response\footnote{On
the order of tens of milliseconds.} to inputs from external
transducers is essential in this application.  Not only must the
response be quick, it must be appropriate for all imaginable driving
conditions.  For example, it would unacceptable for the microprocessor
to interfere with normal braking operations at slow speeds.

Most real-time programming is done in assembly language or a
programming language like C (or C++) that allows the programmer to
explicitly manipulate the contents of specific memory locations.  This
is because communication with a microprocessor is often done through
fixed memory locations. The real-time microprocessor itself is often
programmed in its native assembly language.  The machine instructions
are placed in the microprocessors memory, which is physically distinct
from the memory system of the host computer, so that the instructions
can be executed autonomously by the real-time processor system.


\section{Real-time simulation}
\label{sec:realtimesim}

A real-time application may consist of a host computer that
communicates with several microprocessors, each of which senses some
physical property (like the temperature in a room, or the speed at
which a wheel is spinning) or controls some device. The host and an
individual microprocessor communicate through a ``register'': a
location in the microprocessor's memory. This location is ``mapped''
onto a fixed location in the host processor's memory, so that when
the host references that memory location, it is really referencing the
microprocessor's register. This is why many real-time applications
need to read and write specific memory locations.

The microprocessor usually has a ``program status word'', or {\em
psw}, whose bits are used to indicate the status of the
microprocessor. For example, one of the bits, say {\tt C}, might be
used as follows: When the host computer is ready to read a value from
the microprocessor's register, it (0) sets bit {\tt C} to 1, (1) waits
until {\tt C} is set to 0 (by the microprocessor, to indicate that a
new value is in its register), and then reads the register. This is
how the host computer and the microprocessor communicate and
synchronize.

Because of its built-in security measures, Java lacks the ability to
manipulate the contents of specific memory addresses explicitly, so it
is not well suited to real-time programming.  However, Java does
provide the tools needed to simulate a real-time computing
environment.  Our real-time simulation 
of a home heating system is an example. In it, we simulate this aspect of
communication using a method {\tt readValue()}.

\section{Simulating a heating system}
\label{sec:heating}

Besides a system output
window, our heating-system simulation has five small windows, each of
which represents one component of the home heating system:
\begin{enumerate}
\item {\bf A clock}. You can see the clock ``ticking'' away.
The period 5000 means that the clock ticks every 5,000 milliseconds,
or every 5 seconds. You can change this to a smaller or larger number
of milliseconds by clicking in that text field, typing a
new number, and pressing button {\tt Read period}. (Minimum period is
100 milliseconds.)

\item {\bf The outside temperature} is initially 32 degrees,
since this is Ithaca.  Typically, a microprocessor would be attached
to a sensor that would detect the outside temperature. In
our simulation, the user changes the outside temperature by typing
a new integer and pressing button {\tt Read tem\-per\-a\-ture}.

\item {\bf The furnace} is switched on or
off by the program as needed.  This window just displays its status.
There is a button for turning output on or off.
Press it; thereafter, at each clock tick, a single line of output is
printed in the system output window, telling you the status of the
furnace, the change in the inside temperature due to the furnace being
on, the change due to the difference in the temperature inside and
outside, the total change for this tick, and finally the new inside
temperature.

\item {\bf The desired temperature} is set initially at 68, rather than 72,
to save energy. Change the desired
temperature, as you change the outside temperature. Go ahead;
raise it if you're cold.

\item {\bf The inside temperature} is initially 60 because the thermostat was
at 55 for the night and we just woke up and changed it
to 68. Don't worry, it will warm up soon. The
inside temperature is simulated by the program; how much it changes at
each clock tick depends on whether the furnace is on and the
difference between the inside and outside temperatures.
\end{enumerate}

Experimenting with this home-heat\-ing simulation.
Become familiar with the five components on the screen.


\section{Threads in Java}
\label{sec:threads}

Typically, several programs may be executing on a computer at one
time. Suppose your laptop is connected to a
printer and a modem. One program on your computer could be printing
something, another one could be faxing a file, a third could be
computing something in the background, and a fourth could be some game that you are
playing. Thus, (at least) four programs are executing at the same time
on your laptop.  Each is called an {\em execution thread}.

There is only one CPU (central processing unit) on your computer, so
the four programs can't execute simultaneously. Instead,
your computer allocates a bit of execution time to each
thread. Your operating system uses a priority
scheme to decide which thread should execute next
and how much time it should get. This switching is so fast and
frequent that it gives the illusion of simultaneous execution.

These threads (of execution) can be independent (e.g. the printer
program and your game don't interact), in which case allocation is
fairly easy.

Sometimes, threads have to communicate with each other and
synchronize in some fashion. For example, one thread may
maintain windows on the screen, so when your Java
program ---which has at least one thread of execution--- draws
something in a window, it communicates with the
thread that maintains the window.

In Java, synchronization is provided using three features.  Suppose
$c$ is a class instance that is a thread. Then, execution of {\tt
c.wait()} in another thread $b$ (say) tells the system to stop
executing thread $b$ until thread $c$ executes a statement {\tt
notifyAll}. In other words, execution of {\tt c\-.noti\-fy\-All} tells
all threads that are waiting on $c$ that they can now continue.

There is also a ``synchronize'' property, which can be attached to
methods or to individual statements; it says that all other processes
must be deterred from executing code in this class until this method
or statement has finished --other processes are locked out.

A thread is an instance of class {\tt Thread}. Here are some of the methods of class
Thread that you can use:
\begin{itemize}
\item public void run(). When a thread is created by implementing interface {\tt Runnable},
which we do, this method is called to start the thread executing.

\item {\tt public static Thread currentThread()}. \\This returns the thread that is
currently executing.

\item {\tt public void start()}. Call this to start executing the thread --this method,
which should not be overridden, calls method run.

\item {\tt public static void sleep(long millis)}. \\Make this thread sleep for millis
milliseconds.

\item {\tt public void interrupt()}. Interrupt this thread, so that another thread can
execute.

\item {\tt public final boolean isAlive()}. = ``this thread is alive --has not died''.
\end{itemize}

There are two ways to create a {\tt Thread}. The first, which we don't use, is to use a
constructor of class {\tt Thread}. 
The other way is to implement interface {\tt Runnable},
which requires the class to have a method {\tt run()}, which is called by the system to
execute the thread whenever your program calls method {\tt start()}. This is the method
we use in the home-heating simulation.

Here's a comment from interface {\tt runnable}: In addition, {\tt Runnable}
 provides the
means for a class to be active while not subclassing {\tt Thread}. 
A class that implements
{\tt Runnable} can run without subclassing {\tt Thread} by instantiating a {\tt Thread}
instance and passing itself in as the target. In most cases, the {\tt Runnable} interface
should be used if you are only planning to override method {\tt run()} and no other
{\tt Thread} methods. This is important because classes should not be 
subclassed unless the
programmer intends on modifying or enhancing the fundamental behavior of the class. 

In our Java simulation, we have six programs ---or six
threads of execution--- executing simultaneously.



\section{The SystemClock}
\label{sec:clock}

File SystemClock.java simulates a clock.  At the end of
the line {\tt public class SystemClock...} appears the phrase {\tt
implements Runnable}. Interface Runnable defines one method, {\tt run()}, which
is called to execute a class instance as a thread. We'll look
at {\tt run} later.

So, each instance of this class is a thread of execution.

The class has two variables that deal with the clock itself: {\tt
counter} contains the number of clock ticks that have happened thus
far; {\tt period} is the number of milliseconds to wait between clock
ticks.

Four fields are components that go in the Frame, which you
see on the screen: two Labels, a Textfield, and a Button.

Look at the constructors for {\tt SystemClock} (including method
{\tt initialize}). We use the default layout manager
for Frames, {\tt BorderLayout}. This layout manager allows us to place
components in five places: North, West, Center, East, and South:

\begin{center}
\begin{tabular} {| c | c | c |}
\hline
\multicolumn{3}{| c |}{North}\\
\hline
West & Center & East\\
\hline
\multicolumn{3}{| c |}{South}\\
\hline
\end{tabular}
\end{center}

In method {\tt initialize}, the four calls on method {\tt
add} place the four components in the Frame.

Two methods in {\tt SystemClock} are fairly obvious: {\tt re\-set} sets
the clock back to zero, and {\tt ticks} is used by other processors to
get the clock's value.

Method {\tt run} is of special significance. The main program (method
{\tt main} of {\tt Thermostat}), calls {\tt cThread\-.start()}, where
{\tt cThread} is the variable that contains the instance of {\tt
SystemClock}. In turn, method {\tt start} will call {\tt run} to start
execution of this thread.

Method {\tt run} is basically a loop in which each iteration does the following:
\begin{itemize}
\item   Make the thread sleep for {\tt period} milliseconds. Then wake up.
\item   Notify all the other threads that may be waiting on this one
        (for example, to read the tick) that the thread is now awake.
\item   Increase the time counter (and display it properly in the Frame).
\end{itemize}

The code to notify the other processes should be performed only when
no other process is referencing variables associated with this thread.
This ``lock out'' is indicated by the phrase {\bf synchronize}. We don't
go into any more detail here on this notion of synchronization.

Finally, take a look at method {\tt action}, which is called when a
button is pressed. There is only one button in the Frame, so there is
no need for code to determine what the action was. The pressing of the
button is processed by (0) reading {\tt TextField} {\tt periodField},
(1) trimming whitespace from both ends of it, (2) converting this {\tt
String} value to an integer, and (3) finally storing the integer in
{\tt period} (if the value is at least 100).

\section{InsideTemperature}
\label{sec:inside}

We discuss just two other classes in this project, because the others
are similar.

First, an instance of class {\tt ClockedFrame} is a thread of
execution that synchronizes with the clock. Thus, within this class,
you see a method {\tt run}, which is a loop that at each iteration
synchronizes with the clock. That is all.

Also, this class is a Frame on the monitor screen. This class is a
superclass of {\tt InsideTemperature} as well as of several other
classes, so an instance of {\tt InsideTem\-per\-a\-ture} is an
executable thread.

Now turn to subclass {\tt InsideTemperature}. Besides some components for
the Frame (two labels), there is only one variable, {\tt temperature},
which contains the current inside temperature in the simulation. The
constructor just constructs the Frame, shows it, and returns; there is
nothing special about it.

Methods {\tt getValue} and {\tt setValue} can be called to
reference and set the inside temperature.




\end{document}