Wireless Communication
Lecture 4 – Trunking and Grade of Service
Text Sections 3.5.2, 3.6,7,8
Trunking:
Sharing a pool of lines or channels between a larger number of users. Statistically, they are not all likely to be on the phone at the same time. But as the # of lines is reduced, an individual user may sometimes be unable to access a free line.
Queue: May be used to hold users waiting for a line and increase the quality of service for the user. The user just waits for (hopefully a short period of time) for a dial tone.
Table 3.3: Quality
of Service or Grade of Service (QOS, GOS)
Erlang = total use of one channel (1 call per hour that lasts an hour)
GOS = typical likelihood that a call will be blocked (Erlang B)
Or that a user will have to wait beyond a certain time (Erlang C)
Set-up time = length of time to allocate a trunked radio channel
Blocked call = call that cannot be completed due to lack of channels = lost call
Holding time = H =average duration of call (in seconds)
Traffic Intensity = A = average channel usage (in Erlangs, dimensionless)
Request Rate = l = average call requests per unit time (calls requests/second)
Number of users = U
Number of available channels = C
Traffic intensity of a single user
Au = l H
Total traffic intensity of the system
A = UAu
Traffic intensity per channel (if evenly distributed throughout the channels)
AC = A / C = U Au / C
This is the traffic that is requested by the system, but may not be carried by the system if the capacity is not large enough.
Maximum capacity = C (Erlangs)
Example:
AMPS Allows GOS = 2% (which means that 2 out of 100 calls will be blocked during busiest hour, 4-8 pm on Thursday and Friday nights)
Blocked Calls Cleared (no queue) (Erlang B)
Blocked
Calls Delayed (Erlang C)
Assume:
1) Calls arrive according to a Poisson distribution (in time)
2) There are an infinite # of users
3) All users, including those that are blocked, may request a channel at any time.
4) Probability of a user occupying a channel is exponentially distributed … longer calls happen exponentially less often (a poor assumption at times when people are using the internet).
5) There are a finite # of channels C
Blocked Calls Cleared = Erlang B (calls that are blocked are dropped, no queue)
Probability of a blocked call
TABLE 3.4 (be sure you can interpret this table)
Trunking efficiency:
How many users a configuration of channels can support.
10 channels with GOS = 0.01 can support 4.46 users
whereas 2 groups of 5 channels (also a total of 10 channels) can support 2x1.36 = 2.72 users
Blocked Calls Delayed = Erlang C (blocked calls are held in a queue and delayed)
Assume also that blocked users don’t hang up.
Equations 3.13-3.19
Probability of a blocked call (delay > 0)
This is greater than the delay for an Erlang B system, because users are waiting in line for the next available line.
Probability a user will wait more than t seconds
Average delay for all calls in a queued system
FIGURES 3.6 and 3.7
Work Examples 3.4-3.7
Cell Splitting (Figure 3.8,9)
Sectoring (Figure 3.10)
Microcell Zone (Figure 3.12)
Work examples 3.8-9