Take a look at the core Encode and Encode::Supported modules, specifically the encode and decode commands. The PerlIO::encoding module calls those commands internally.

Am I going to need to swap the bytes on every character I read in?

No way. Encode is set up to handle both Big-Endian and Little-Endian. You're having a problem with the Byte Order Mark because you specify an encoding of ucs2, which is an alias for the big-endian format UCS-2BE, but your BOM indicates that the data from the stream is in little-endian format. Perl doesn't like that, and it complains.

Here's a script which illustrates that what you want to do is possible:

#!/usr/bin/perl
use strict;
use warnings;

binmode( STDOUT, ':utf8' );

my $big_endian_BOM = pack( 'n', 0xFEFF );
my $smiley         = pack( 'n', 0x263A );
my $ack            = pack( 'n', 0x0006 );
my $newline        = pack( 'n', 0x000A );

my $data = $big_endian_BOM;
$data .= "$smiley$newline$ack$smiley$newline$ack";

open( my $fh, '<:encoding(ucs2)', \$data ) or die $!;

local $/ = $ack; 
while (<$fh>) {
    chomp;  
    print;  
}
[download]

It prints two smileys, each on its own line. No errors or warnings.