BASICS OF QUALITY OF SERVICE IN WIRELESS NETWORKS 13
probability. For example, multimedia playback application can choose delay requirement d
req
for
multimedia flow’s packets with
d
req
= min (d D | D = {d
1
, d
2
, ..., d
N
} and Prob (d [d
min
, d
max
]) > 95%), (1.2)
where D is the set of multiple delay levels, given by the application. Often, QoS-aware resource
functions, services, and protocols work with statistical average values and estimates of QoS param-
eters. For example, if a QoS operation measures N round-trip delay measurements (d
1
, d
2
, …, d
N
),
then it can use
avg
1
1
2
N
i
i
d d
N
=
=
Σ
, as the average value to estimate end-to-end delay [8]. Another
example can be the loss rate estimate for flow f : L( f ) = 1 - j /k, where k is the number of packets a
flow attempts to transmit in a time interval T, and j is the number of attempts that result in success-
ful packet transmission, while k - j packets are dropped by the MAC protocol [9].
1.4 FUNCTIONAL COMPONENTS TOWARD QOS
PROVISIONING
As we discussed, above, due to limited wireless resources and other challenging characteristics of
wireless networks, to achieve QoS is hard. If the service provider aims to deliver and achieve statisti-
cal or proportional QoS requirements, given by a user, it is necessary to include certain fundamental
functions within the resource management and protocols of wireless networks to provide these QoS
guarantees. We will discuss four major necessary functional components: (1) admission control, (2)
QoS enforcement via scheduling, bandwidth management, rate control, and incentives, (3) QoS adap-
tation via estimation and feedback control, and (4) QoS-aware routing. Depending on the level and
strictness of the QoS requirements, the wireless service providers might provide all four QoS-aware
functions in the protocol stack or just three QoS-aware functions (scheduling, adaptation, and rout-
ing) or just two QoS-aware functions (scheduling and adaptation) or just one QoS-aware function
(adaptation). Using all four QoS-aware functions, we get a very pro-active approach and get closest
with the network performance to the desired QoS requirements. However, this is also the most
expensive approach. Using three, two, or one QoS-aware functions, we get reactive approaches, since
we do not perform any admission control at the entrance of a QoS-sensitive flow and do not keep
any state per flow. What we do is react to the resources availability and adapt each packet of a flow to
the new resource situation. In this case, we must monitor the network performance of all flows, and
based on their performance over given resources availability, we then adapt/determine the needed
priorities and QoS level for scheduling, choice of a service class, and routing, respectively. The less
QoS-aware functions we deploy in the wireless protocol stack, the less costly the wireless solutions
will be, but the farther we get from the conformance of user desired QoS requirements.

Get Quality of Service in Wireless Networks Over Unlicensed Spectrum now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.