Of course, if you're always looking up data by MAC, it could make more sense to structure it as a giant hash, keyed by the MAC, of hash (refs) of the data. That would make looking up particular MAC's faster and a bit cleaner. It may be easier to read too, depending on your own preferences.

#!/usr/bin/env perl5
use strict;
use warnings;

my %clients;
my $curmac;

while(<DATA>)
{
    chomp;

    # Assume a format of "Type Data"
    my ($type, $data) = split / /;

    # If it's a MAC address, assume later lines refer to this MAC
    if($type eq "MAC")
    {
        $curmac = $data;
        next;
    }

    # RSSI and SNR just get saved
    $clients{$curmac}{RSSI} = $data if $type eq "RSSI";
    $clients{$curmac}{SNR} = $data  if $type eq "SNR";
}

# Show it
use Data::Dumper; print Dumper \%clients;

# Find a given entry
sub showmacsnr
{
    my $mac = shift;
    my $ent = $clients{$mac};

    if($ent)
    {
        print "MAC $mac has SNR $ent->{SNR}\n";
    }
    else
    {
        print "Can't find MAC $mac\n";
    }
}

showmacsnr("ba:98:76:54:32:10");
showmacsnr("thi:is:is:nt:re:al");


# Sample data
__DATA__
MAC 01:23:45:67:89:ab
RSSI 12
SNR 18
MAC ba:98:76:54:32:10
RSSI 7
SNR 3
[download]

% ./tst.pl
$VAR1 = {
          'ba:98:76:54:32:10' => {
                                   'SNR' => '3',
                                   'RSSI' => '7'
                                 },
          '01:23:45:67:89:ab' => {
                                   'SNR' => '18',
                                   'RSSI' => '12'
                                 }
        };
MAC ba:98:76:54:32:10 has SNR 3
Can't find MAC thi:is:is:nt:re:al
[download]