If you are reading telemetry from one port and writing (derived?) waypoints to the other, why do you need concurrency? On the surface, it sounds much like a simple filter program using sockets instead of pipes?

Something as simple as this might be all you need:

use IO::Socket;
use XML::Simple;

my $cTelemetry = IO::Socket::INET->new( 'localhost:12345' ) or die $!;
my $cWaypoints = IO::Socket::INET->new( 'localhost:6789'  ) or die $!;

while( <cTelemetry> ) {
    my $xmlTelemetry = XMLin( $_ );
    if( my $xmlWaypoint = deriveWaypoint( $xmlTelemetry ) ){
         print $cWaypoints $xmlTelemetry;
    }
}
[download]

That leaves you to write the code for deriving waypoints from telemetry data, but unless that is a slow process, there does not seem much scope for concurrency here?