AFP OPTIMIZER BASICS

By Vinod Rathod

1 © Nokia Solutions and Networks 2014

1. OPTIMIZER BASICS

NetAct Optimizer is used in the statistical and/or network wide optimization process in NetAct. Optimizer provides visibility to
current network behavior by combining the actual GSM network configuration parameters and measured performance statistics
with advanced visualization and analysis functionality. Parameters can be optimized manually for small changes or automatically
by choosing from the range of optimization solutions provided by Optimizer. Optimizer can be used for a single cell or for a
whole region. The result of optimization algorithms can be visualized on a geographical map before downloading the
optimization plan to the network. The plan with the changed parameters is sent to RAC where it is validated and provisioned to
the network. Some of the basic functionalities of Optimizer includes

• Optimizer uses statistical performance measurement data.

As the input data for algorithms is accurate (measurements of a real network), the output is also more accurate than with a
signal propagation estimate based process in Asset.

• Using measurements makes the tuning process faster.

Instead of heavy raster map based calculations - where, for example, the interference matrix is calculated by considering signal
strengths in each map pixel - a mobile measurement report is used. When the data is processed in Optimizer, only some analysis
is needed.

• Increased level of automation.

With Optimizer, the whole optimization cycle is faster than with planning tools. As Optimizer is implemented in the NetAct
Framework, the actual configuration data and measurement reports are available for processing. The network topology in
Optimizer is always consistent with the actual network data. When running Optimizer for the first time, some customizing is
needed, such as parameters needed to guide the generation algorithms. Once the parameters are set, the next optimization
2 © Nokia Solutions and Networks 2014
round is more effortless.
 2. MEASUREMENT BASED ANALYSIS
Interference Matrix, which is primarily based on interference related measurements but can also be
complemented with predictions, is the main core of Optimizer. Interference matrix represents the
interference relations between cells. Interference can be computed or expressed with different
mathematical methods such as ARP (Average Received Power), CIP (Carrier over Interferer Probability) or FEP
(Frame Erasure Probability). Interference in Optimizer is primarily computed based on measurements that
have been made in an operated network. However in case measurements are not satisfactory because some
of missing data (for example, as in the case of new objects, or blind spots), it is possible to use predicted
interference values to complement measurements. This is done by combining data from several sets and by
defining a priority for the sets to be used.

3 © Nokia Solutions and Networks 2014

3. DIFFERENCE BETWEEN PREDICTIONS & MEASURMENT BASED TOOLS

Interference Matrix is generated in both predictions and measurement based tools. This interference matrix is then used in the
calculation of cell interference from other cells. Optimizer generates the interference matrix based on following measurements
• CIP (Carrier to Interference Probability)
• ARP (Average Received Power)
• FEP (Frame Erasure Probability)

Carrier over Interferer Probability (CIP) is a probability of a random call experiencing interference that exceeds a given
threshold. Usually, an adequate threshold for good quality calls is considered to be about 9 dB. CIP is represented as a probability
factor having floating point numbers with four number accuracy (two decimals). The interval is [0.01,100.0] and unit percentage
(%).

The Average Received Power (ARP) is a measurement for interference or overlap between any two cells. It is simply the relative
rx level measured from one cell by mobiles served by another cell. ARP is represented as floating point numbers with four
number accuracy (two decimals). The interval is [0.0,63.0] and unit RXLEV.

Frame Erasure Probability (FEP) is an estimated probability for a speech frame to be erased. FEP is represented as a probability
factor having floating point numbers with four number accuracy (two decimals). The interval is [0.01,100.0] and unit percentage
(%).

Interference Matrix in Prediction Based Tools is calculated with the help of predictions. The predictions usually depend upon
propagation models. The model describes the behavior of RF waves in certain environment and clutter. The Interference matrix
generated
4 with©the
Nokiahelp of predictions
Solutions will also be used for determining the interference values between cells but this value will
and Networks 2014
not be the real value.
4. Inter Cell Dependency Matrix (ICDM)

Development of an efficient frequency planning algorithm requires in-depth knowledge of the cellular environment
and live system measurements can also be of great help in this regard. As the ICDM is calculated by taking live network
data and system measurements, it provides a better approach to address the frequency allocation problem. Area and
traffic based inter cell dependency matrices are the two types of ICDMs that are given as input to the frequency
planning algorithm. The calculation of best server for each bin in the coverage area provides the foundation for
calculating the area based ICDM. Area based ICDM contains the percentage of area in a cell that is affected by another
cell if those two cells are given either co-channel or adjacent channel frequency as shown in fig. 1 which contains the
percentage interfered area of a particular cell due to another cell

The diagonal elements in the generated ICDM have to be

zero because a base station can obviously not interfere in
its own best server region. Percentage interfered area
basically shows the percentage area of the cell in which
the interferer cell has a signal strength within the range of
9dB and where the best server power is not 9dB stronger
than the interferer power as shown in fig. 2.

5 © Nokia Solutions and Networks 2014

 In fig. 2, considering cell A as the best server, intersection
area between cell A and B shows that area of cell A where
its power is not 9dB stronger than that of cell B and has
resultantly become interfered area and this percentage of
interfered area between cell A and cell B is stored in the
area based ICDM.
Although area based ICDM shows all the potential
interferers in each base station coverage area, it cannot
accurately describe the actual traffic profile.

The area based ICDM assumes uniform traffic for each bin. For real time systems, this is rarely the case.
Hence we need traffic based ICDM from live networks that show the actual traffic profile in the
geographic area. Traffic based ICDM provides the basis for the automatic frequency planning in the
network because it is based on live data of the network and provides information that is authentic in
nature in the form of PIT, the percentage interfered traffic. PIT is the percentage of traffic for a cell that is
affected by another cell operating at same or adjacent frequency as shown in fig. 3, which contains the
correlation between any two cells in the form of PIT.

6 © Nokia Solutions and Networks 2014

 Apart from the diagonal elements there are other zeros in
the traffic based ICDM which are basically due to two
factors:
1). If base stations A and B are sharing same frequency F
but they are located far apart as shown in fig. 4 then there
will be no interfered traffic between them resulting in a zero
entry in the ICDM.

1). If base stations A and B are sharing same frequency F

but they are located far apart as shown in fig. 4 then there
will be no interfered traffic between them resulting in a zero
entry in the ICDM.

2). If base stations A and B are operating at the same

frequency F in each other’s vicinity as shown in fig. 5 then
there must be interfered traffic between them.
7 © Nokia Solutions and Networks 2014
But, in the calculation of traffic based ICDM the mobile station basically acts as a probe for the measurement
of neighboring frequencies and thus does not account for the interferers operating on similar frequency. This
is an inherent limitation in the traffic based ICDM. To prevail over this limitation, the area based ICDM has
been incorporated into the traffic based ICDM. The weightage given to percentage interfered area is less as
compared to the percentage interfered traffic. Now the final matrix obtained after the incorporation of both
traffic based ICDM (with 80% weightage) and area based ICDM (with 20% weightage) as shown in fig. 6 may
also contain values which will be zero but these ‘zero’ values in the matrix correspond to the fact that base
stations are located far apart.

8 © Nokia Solutions and Networks 2014

5. Concept of Cost and Rule Breakage
The concept of cost is very important in the optimization process and its magnitude basically governs the
assignment of frequencies thus acting as the basic judgment criteria for any optimal frequency plan. It is the
penalty encountered whenever an unwanted frequency assignment takes place. Events, which are responsible for
incurring cost, are called rule breakages. So, strictly following these rules will lead to an optimized plan with
minimum cost.
• Strict rule says that co or adjacent frequency must not be assigned to co-sited cells and no violation of this rule
should take place. Cost incurred on the breakage of this rule is very large in magnitude.
• Moderate rule says that co or adjacent frequencies must not be assigned to any of the remaining interferers.
Violation of this rule should also be avoided but its breakage incurs lesser cost as compared to strict rule
breakage.
For a cellular network comprising of ‘N’ cells based on the strict and moderate rules with optimum frequency
allocation, the optimum cost function defined in equation (1) is the combined ICDM percentage interference of
interferer cell ‘i’ with interfered cell ‘j’ according to their co-frequency and adjacent frequency relation [4].

9 © Nokia Solutions and Networks 2014

10 © Nokia Solutions and Networks 2014
REFERENCES
[1] Nokia System Description of the NetAct Optimizer NETACT Detail 1.6
[2] Product presentation NetAct Optimizer
[3] Natasa Neskovic, Aleksandar Neskovic, Dorde Paunovic “Automatic Frequency Planning Algorithm in a Real
Land Mobile Radio System.
[4] Lee, W.C.Y., Mobile Communications Design Fundamentals, John Wiley & Sons, inc, 1993.
[5] Mishra, A.R.: Radio Network Planning and Optimization. In: Fundamentals of Cellular Network Planning and
Optimization: 2G/2.5G/3G... Evolution to 4G. Wiley (2004) 21–54
[6] Frequency Planning SOP by Kamran Afzal LCC Organization.

11 © Nokia Solutions and Networks 2014