In wireless communication networks, performance goals are often conflicting
with each other. For example, in point-to-point multi-input-multi-output
(MIMO) links, spectrum efficiency and reliability are in a tradeoff
relation; in ad-hoc wireless networks, we need to sacrifice throughput
to decrease packet delay. In this dissertation, we investigate problems
on the diversity-multiplexing tradeoff (DMT) in cellular wireless
communication networks and problems on the throughput-delay tradeoff in
ad-hoc wireless networks.
We first consider two topics on the DMT in cellular uplinks
(or multiple-access channels, MACs):
(1) the diversity-multiplexing-delay tradeoff (DMDT) in
a random-access scenario and (2) an explicit construction of space-time
coding scheme achieving the DMT. For the random-access scenario, we propose
an incremental-redundancy automatic repeat request (IR-ARQ) scheme.
We prove that our scheme successfully exploits both ARQ diversity and
joint-decoding advantage, and achieves a better DMT than other
existing protocols, such as Tsatsanis et al.’s network-assisted
diversity multiple-access (NDMA) and Gallager tree algorithm.
Next, we propose a lattice-space time (LAST) coding/decoding scheme
in MACs and prove that it achieves the optimal
DMT in MACs. Although our result is established using a random coding
argument, it is important to note that the proposed scheme is
explicit in a sense that it lends itself to a structured encoder
and an efficient decoder which does not require exhaustive search.
Next, we formulate and analyze the DMT in delay-constrained cellular
downlinks (or broadcast channels, BCs). We show that dirty-paper
precoding achieves the optimal DMT in BCs. Furthermore, we analyze the
DMTs for a few suboptimal precoding schemes. In particular, we find that
vector precoding schemes such as vector perturbation of Peel
et al. and LLL lattice reduction achieve the optimal DMT of
a type of BCs, in which a BS has a larger number of antennas than
the number of users each with a single antenna.
Besides the DMTs in cellular wireless networks, we also analyze the
throughput-delay scaling in ad-hoc wireless networks in which Ozgur et al.’s hierarchical cooperation is allowed.
We propose a hierarchical multihop scheme for these networks, and prove
that the throughput-delay relation of the proposed scheme is, for
any small ε > 0,
D(n) = Θ(nεT(n)) up to T(n)=Θ(n1-ε),
where T(n) is the aggregate throughput and D(n) is the packet delay
in such a network with n nodes.
Thus our result successfully extends the throughput-delay
result of El Gamal et al.’s where they proved the same
throughput-delay relation using a multi-hop scheme, only up to
T(n)=Θ(n1/2).