SQL & PL/SQL :: Cross Join Query To Remove Repeating Combination

I have constructed a cross join query, with the test case below

create table ajit_sites (
site_id char(1));
insert into ajit_sites values ('A');
insert into ajit_sites values ('B');
insert into ajit_sites values ('C');
COMMIT;

sql below is constructed to display combination of all sites (cross-join), it also removes records where "origin" is the same with "dest"

select
a.site_id origin, b.site_id dest
from
(select site_id from ajit_sites) a,
(select site_id from ajit_sites) b
where
a.site_id <> b.site_id b

Is there any way i could remove records with the behavior below

Origin , Dest
A , B
B , A

For instance from the example above, i want to only retain one of the records since record (A, B) or record (B, A) means the same.