A common approach is to submit an extended version of the paper to arXiv containing all the additional material needed (e.g., full proofs), referencing the arXiv paper from the submitted conference paper for further details as necessary.
If that does not sound like a good idea, it might suggest that the paper is better submitted to another venue, or perhaps directly to a journal.
Edit: Just looking now specifically at INFOCOM requirements, peer review is double-blind, which means publishing on arXiv becomes more complicated (obviously you cannot reference an extended paper where the authors are given). What may perhaps work is to informally publish an extended version of the paper online (in a "neutral" location of course, not giving away the identity/affiliation of the author(s)) and reference that in the submitted version, then if the paper is accepted, consider publishing the arXiV version with author names and reference that from the camera ready version. Admittedly this gets more complex, where you should probably contact the PC chairs of the specific conference to see what it the norm in such cases.
Searching further in the case of INFOCOM:
- Papers with the same title and abstract should not be posted on a public website, such as arxiv.org, or transmitted via public mailing lists.
- We allow the use of an anonymous Microsoft OneDrive link for the reviewers to download an extended version of the submitted manuscript with details of mathematical proofs. Needless to say, the extended version of the submitted manuscript will need to be anonymized, following the guidelines above. It is the responsibility of the authors to make the OneDrive link (and its content) anonymous, with respect to both the owner of the link and people who visit the link. The reviewers, however, are not obliged to read the extended manuscript when evaluating a submitted paper.