I use OPNET software for simulating my projects (OPNET
Modeler). OPNET Modeler is event driven simulation tool, and is very suitable
for my project.
Modeler’s object-oriented modeling approach and graphical editors mirror
the structure of actual networks and network components. Modeler supports
all network types and technologies. OPNET Modeler is based on a series of
hierarchical editors (network, node, and process). At node level, I will
create all devices that I need for my simple wireless network (mobiles and
base station). The network device such as my base station consists of functional
elements called modules at node level.
Inside each module I will develop process model which will specify module's
behavior.
Currently I am working on Call Admission Control (CAC) scheme for 3rd Generation
wireless networks.
Dynamic Prioritized CAC (DP-CAC) is proposed here. It
is implemented at base station (BS) on a frame-by-frame basis. Admission
requests received on previous frame’s random access channel (RACH) are collected
at BS. The differentiation between handover and new calls is then made by
prioritization function, as well as differentiation of various traffic classes
supported in UTRA-TDD network. Prioritization function as discussed below
ensures that highest priority is given to handover calls, because from users’
point of view it is worse to lose already active call (call dropping), then
to block new call. New calls are only considered for admission if dropping
probabilities of higher priority classes (as defined below) are under acceptable
values (thresholds). Dropping rates are calculated dynamically by the system.
Furthermore different traffic classes are differentiated by using different
thresholds. Threshold values are adjusted dynamically by the system in order
to respond to changing traffic conditions. If the call is not dropped/blocked
in the prioritization phase (i.e. it passes prioritizer function), a simple
resource check is performed to find out if there are enough resources to
admit a call. Admission decision is based on guaranteeing QoS requirements
for each call at each time slot of TD-CDMA system for all admitted users.
The available bandwidth is divided into circuit switched dedicated channels,
and packet switched shared channels. The dedicated channels are assigned
to highest priority RT calls, and their QoS requirements are guaranteed in
a deterministic sense. On the other hand, high priority NRT users (and RT
users when dedicated channels are not available) are admitted if their QoS
is guaranteed in a statistical sense based on calculations by packet scheduler.
In this way, performance of packet scheduler is integrated in the admission
decision for packet switched shared channels.
Simulation Environment
To evaluate performance of the proposed CAC scheme,
a simulation model using event driven simulator OPNET is developed. A multiple
cell environment consisting of small macro cells (omni directional coverage)
is considered in the model. The system consists of a target cell with a single
base station (BS) where a CAC algorithm is implemented. Mobile stations (MS)
seeking admission are located either in the target cell (ones initiating
new connections), or they are entering the target cell (handover or ongoing
connections) from 6 neighboring cells (1st tier only). The system is shown
below.
Mobiles send admission request to BS, which implements the CAC algorithm
and assigns resources accordingly.
Network Topology and Simulation Parameters
Handover admission requests are modeled as two sources.
Source H1 generates RT handover calls coming from neighboring cells. Their
arrivals are modeled as Poisson random process with mean arrival rate
(calls/second) ( ). Source H2 generates handover calls of NRT traffic classes
(interactive and background in UMTS). Their arrivals are also modeled as
Poisson random process with mean arrival rate ( ).
New calls (ones originating from the target cell) are modeled as two sources.
Source N1 models new calls of RT traffic classes. The arrivals are assumed
to be Poisson with mean arrival rate (calls/seconds). Similarly
new calls initiating NRT traffic are modeled as a Poisson source N2 with
mean arrival rate of (calls/second).
Traffic sources are implemented using existing OPNET generator process
models. The process model for proposed CAC algorithm is developed inside
BS node (bs_cac_control module in the figure below).
Topology and snapshot of the model (at the OPNET’s node level) is shown below:
Inside BS I developed process model which defines BS's behavior. The process
editor uses a powerful finite state machine (FSM) approach to support specification,
at any level of detail. A sample from process editor is shown below:
The air interface is TD-CDMA (TDD UTRA). The available channels (time-code
pairs) are divided into dedicated circuit switched physical channels (DPDCH
as specified in UMTS) used for RT connections, and to shared packet switched
channels (DSCH in UMTS specifications) used for NRT and RT connections, and
managed by packet scheduler (as shown below).
Once a RT connection is assigned dedicated circuit switched channel(s), it
is held for the duration of a call/session. The dedicated channel holding
time is assumed to be exponentially distributed random variable with mean
seconds. Perfect power control is assumed, so that the number of such dedicated
channels (time-code pairs) in one frame corresponding to lowest data rate
is fixed, and set to 100 in simulation. Multirate concept is implemented
using multicode. The bit rate of a single channel is bps. If a source
needs to transmit at rate bps, it needs channels (time-code pairs).
Packet switched shared channels are managed by packet scheduler, and they
are shared on a frame-by-frame basis.
The TD-CDMA frame consists of 15 time slots. The frame duration as specified
in 3GPP UMTS standard is ms.