
284 Advances in Mobile Cloud Computing Systems
8.4.3 Maximizing TCP Throughput with Delayed CSI for Mobile
Cloud Services
At time slot t, the system state S
t
is an unobserved random variable.
The wireless network cloud selects the cooperating RRHs and allo-
cates the rate for MCC users, denoted as a
t
. Denote the end-to-end
throughput of a mobile user u given by Equation (8.15) as a random
variable η
u
(a
t
,S
t
), then
B
u=1
η
u
(a
t
,S
t
) is the sum throughput of the
system. The number of time slots considered is h, which is called
the number of horizons in Markov decision process literature [17].
The cumulative rewards over h horizons is
t=h
t=1
u=B
u=1
η
u
(a
t
,S
t
).