QoS

Quality of service is a term used by many and has also the same
amount of meanings. For this discussion, QoS is a description of the
bear channels capability for delivering a particular grade of service,
GoS. The GoS is typically defined as a blocking criteria, and for circuit
switched data, it is defined by Erlang B, Erlang C, or Poisson
equations. For packet data, the relationship for QoS/GoS is blocking,
Erlang C, and delay, to mention three of the key attributes.
With the introduction of packet data with CDMA2000, the traffic modeling
for packet-switched data involves the interaction of the following items:
■ Number of packet bursts per packet session
■ Size of packets
■ Arrival time of packet burst within a packet sessions
■ Arrival times for different packet sessions
The packet usage is relatively an unknown area for wireless mobility
systems on a mass-market basis. The issue of where, when, and how much
do you dimension a system for packet data will always be a debate between
marketing and technical teams. However, in light of the fact that packet
data usage is at its infancy, there is little guidance from which to go forth
and design the network from. However, ITU-R M.1390, which gives a
methodology for calculating the spectrum requirements for IMT-2000, has
some guidelines for data dimensioning and the following tables are
extracted from that specification. The values in the tables should be used as
a guide to establish packet loading for dimensioning when market specific
data is not available for a numerous amount of reasons.
It is important to note that all the services defined previously in
Table 13-6 and elaborated on in Table 13-8, are either symmetrical or asymmetrical.
Of the services listed previously, only MMM and HMM are asymmetrical;
the rest are symmetrical service offerings.

Note that the penetration rates shown in Table 13-9 for the services are
the same and come to more than 100 percent for any location. It is also important
to note that the previous numbers indicate that 73 percent of the system
usage is expected to be voice-oriented; 10.8 percent is for SM, 3.51 percent for
both SD, MMM, and HMM, whereas 6.75 percent for HIMM. However, the
following tables will help provide additional insight into possible traffic
dimensioning requirements. (See Tables 13-8, 13-9, 13-10, 13-11, and 13-12.)
The next step is to determine the traffic forecast of user by service type.
The method for achieving this value is determined by the equation for each
of the service types and locations defined, building/pedestrian/vehicular.
Traffic/user
BHCA  call duration  activity factor
0.9  120  0.5

54 call sec during the system busy hour for downlink or uplink voice
service for a building environment.
The amount of circuits required for circuit switched voice, switched data,
and HIMM services is determined via Erlang B, although the remaining
packet data services are determined via Erlang C.
Now the next question is to define the next set of variables that need to
be established to help dimension the rest of the packet network. For symmetrical
services, the dimensioning is straight forward, well as straight as
it can get. However, for asymmetrical service, a few more details are
required that are used for the selection and performance of the PDSN:
Transmission time (s)
NPCPS  NPPPC  NBPP
 8 bits per byte/1024 Kpbs
Total session time
packet transmission  [(PCIT  (NPCPS-1)]
 [PIT  (NPPPC-1)]
Activity factor
packet transmission time/total session time
The data used for uplink and downlink calculations is extracted from
Table 13-13. The results are then entered into the traffic calculation section
discussed later.