I, personally, have never see any authors in IEEE transactions listed as "co-first" authors, as such
John Smith†, Bill Lee†, and Boss James
...
† These two authors contribute equally to the work.
Is this allowed in IEEE transactions?
Yes, it is allowed. Google scholar is your friend in such cases.
I employed this query, which looks for publications with "IEEE Transactions" in them, the text "contributed equally to", and one in (text | paper | work | article).
The results show several papers published in IEEE Transactions, which specify multiple equal co-authorship. The way this is specified seems to be non standardized.
For example, http://dx.doi.org/10.1109/TNB.2006.875054 specifies that "The first two authors contributed equally to this work." in the space reserved for details such as when the manuscript was received and revised.
In http://dx.doi.org/10.1109/TED.2009.2034804, the equal contribution is specified in the ACKNOWLEDGEMENT section of the paper.
Both ways look good to me. The first case is likely achieved by communicating the authors' contribution in the cover letter / space for staff communication. The second case is achieved by simply writing the contribution in the text body.
Some people use alphabetic sorting of authors.
Assuming that the first author contributed "most" is a very fragile and questionable approach.
This is not even consistent within a field.
It's largely the personal preference of the supervisor, usually.
I have never seen co-first authors either ... you can use alphabetic ordering to indicate more-or-less equal contributions, or you can use partial alphabetic ordering (the first group is alphabetically ordered and the last group is not), or non-alphabetic to indicate that the first author contributed the most.
Ultimately, the system only works when co-authors are honest to the public about their contribution. If you're worried that co-authors will cheat you out of credit, you probably should not be working with them.