A first step toward distributed scheduling policies in cellular ad hoc networks

The future of cellular wireless networks could see multihop transmission through relays as a means of improving performance of existing cellular systems. Incorporating intermediate relays is the first step in an evolution of wide area wireless information networks (see Pahlavan, K. and Levesque, A. H., 1995; Naghshineh, M., 1999) from the extreme of the cellular paradigm to the opposite extreme of ad hoc. For this "cellular ad hoc" network, we simulate and compare two downlink scheduling policies. We consider a purely centralized algorithm (T&E) (see Tassiulas, L. and Ephremides, A., IEEE Trans. Auto. Control, vol.37, no.12, p.1936-48, 1992), which achieves notable throughput gains, as well as a more practical hybrid algorithm (HY) one step away from T&E in the direction of being distributed. Naturally, a more practical algorithm would be expected to exhibit worse performance. However, although HY is interference-limited when lower powered relays are used, HY offers comparable performance to T&E, and, for asymmetric user distributions, HY even outperforms T&E.