Monday, September 7, 2009

TCP/IP Socket Communications in MATLAB - Part 2

So my little blog post about TCP/IP Socket Communications in MATLAB was rather popular despite being a very simple example.

One of the main limitations that people found when trying to utilise the server/client scripts for their own applications, was that it was incredibly inefficient at shifting large volumes of data around. This is thanks to the following lines:
message = zeros(1, bytes_available, 'uint8');
for i = 1:bytes_available
message(i) = d_input_stream.
readByte;
end
For each individual byte within the message, there was the overhead of a function call.

I mentioned in comments on MATLAB Central that I made a Java class that bypassed this overhead and allowed for more efficient transfer. Being in a caring sharing mood (and getting sick of emailing it to people all the time!) I thought I would upload it also.

The is available on the Mathworks File Exchange: TCP/IP Socket Comms Example using Java Class

Compiled Java?

The client/server scripts are essentially identical to their previous iterations.

The server still uses a ServerSocket and the resulting DataOutputStream to which we write data.

The client side uses a Socket to connect to the specified host and port which provides us an InputStream which we wrap in a DataInputStream to read data from. However, instead of using the interface of DataInputStream directly, we hand off to another function (the new DataReader class) to perform the read.

The code for the example client is outlined below. I have bolded the changes. (The server is unchanged).

client.m
% CLIENT connect to a server and read a message
%
% Usage - message = client(host, port, number_of_retries)
function message = client(host, port, number_of_retries)

import java.net.Socket
import java.io.*

if (
nargin < number_of_retries =" 20;" style="color: rgb(0, 153, 0);">% set to -1 for infinite
end

retry = 0;
input_socket = [];
message = [];

while true

retry = retry + 1;
if ((number_of_retries > 0) && (retry > number_of_retries))
fprintf(1, 'Too many retries\n');
break;
end

try
fprintf(1, 'Retry %d connecting to %s:%d\n', ...
retry, host, port);

% throws if unable to connect
input_socket = Socket(host, port);

% get a buffered data input stream from the socket
input_stream = input_socket.
getInputStream;
d_input_stream =
DataInputStream(input_stream);

fprintf(1, 'Connected to server\n');

% read data from the socket - wait a short time first
pause(0.5);
bytes_available = input_stream.available;
fprintf(1, 'Reading %d bytes\n', bytes_available);

data_reader = DataReader(d_input_stream);
message = data_reader.readBuffer(bytes_available);


message = char(message'); % Data comes out as a column vector

% cleanup
input_socket.close;
break;

catch
if ~
isempty(input_socket)
input_socket.close;
end

% pause before retrying
pause(1);
end
end
end

Instead of looping for each byte, we now ask the DataReader object to read the specified number of bytes and return them to us. So a single function call per read. This could be used to read buffers of expected data sizes.

DataReader

Below is the Java source for the DataReader class (available in the MATLAB File Exchange also).

import java.io.*;

class DataReader
{
public DataReader(DataInput data_input)
{
m_data_input = data_input;
}

public byte[] readBuffer(int length)
{
byte[] buffer = new byte[length];

try
{
m_data_input.readFully(buffer, 0, length);
}

catch (StreamCorruptedException e)
{
System.out.println("Stream Corrupted Exception Occured");
buffer = new byte[0];
}
catch (EOFException e)
{
System.out.println("EOF Reached");
buffer = new byte[0];
}
catch (IOException e)
{
System.out.println("IO Exception Occured");
buffer = new byte[0];
}

return buffer;
}

private DataInput m_data_input;
}
It is very simple class that is given a DataInput to read from, and has a single public method readBuffer(int length) that returns a byte array.

If you have the Java Software Development Kit installed, you can compile the class with:
C:\matlab\matlab_socket> javac data_reader.java

Running the Client/Server example with DataReader

This is the same as the previous example, with one exception. You must tell MATLAB where to find the compiled DataReader class (In my case it is located in the C:\matlab\matlab_socket directory):
>> javaaddpath('C:\matlab\matlab_socket');
Opening up two instances of Matlab:
% Instance 1
>> message = char(mod(1:1000, 255)+1);
>> server(message, 3000, 10)
Try 1 waiting for client to connect to this host on port : 3000
Try 2 waiting for client to connect to this host on port : 3000
Try 3 waiting for client to connect to this host on port : 3000
Try 4 waiting for client to connect to this host on port : 3000
Client connected
Writing 1000 bytes

% Instance 2 (simultaneously)
% NOTE: If the 'server' was runnning on a non local machine, substitute its IP address
% or host name here:
% data = client('10.61.1.200', 2666); % To connect to server at IP 10.61.1.200:2666
>> javaaddpath('C:\matlab\matlab_socket');
>> data = client('
localhost', 3000)
Retry 1 connecting to
localhost:3000
Connected to server
Reading 1000 bytes

data =



 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~     

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~     

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~     

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~    

Why not call readFully() from within client.m?

The more astute readers may be asking the above question. Why not do something like this in client.m:
message = zeros(1, bytes_available, 'uint8');
d_input_stream.
readFully(message, 0 bytes_available);
At first glance that seems great. It runs without error, but hey, you don't seem to be getting any output?¿?

The problem is that DataInputStream.readFully() takes a reference to a byte array to populate.

When you pass things around in MATLAB that are to be modified, MATLAB supplies a copy. Thus the Java most likely sees a byte array reference, and populates it, but it is only the copy that is populated, not the original message array.

Until there exists some way in MATLAB to pass references to Java methods, I am stuck using helper Java classes (If anyone else has ideas on how this could be done better, feel free to comment!).