Radio Network Optimisation

Optimisation of the network, is done to check the performance of the network, just
after it is made operational and to get best possible quality of service.

Objective:
The objective of optimisation procedure is: a) To check whether the network meets the customers given requirements, on the basis of which network was designed. b) To check whether the parameters and configurations are defined correctly or not. c) To find out and suggest changes in the defined parameters and configurations to achieve best possible quality of service.
Quality of service can be characterised by factors such as coverage, accessibility to the

network, speech quality and number of dropped and blocked calls. A number of parameters are checked as a measure of quality of service by using a drive test system. Drive Test system comprises of a test mobile phone, software to control and log data from the phone and a Global positioning system receiver for position information. A drive test system can only indicate the type of problem in the network that exists, it doesnt indicate cause of the problem but with the help of knowledge of possible reasons of a problem, one can trace the cause.

Following steps are taken to fulfil the objective of network optimisation using a drive test tool:
a) Collection of Data and extraction of relevant information from it. b) Analysis of the extracted data. c) Suggesting changes in the network configurations based on the analysis.

Collection of Data and extraction of relevant information:

Drive test involves setting up a call to best carrier and driving along the roads. While driving the radio parameters and air interference signal data are collected as a log file. In general following parameters are checked during the drive test for different categories of terrains like dense urban, sub-urban, rural, highways and for different clutters like in building, residential areas, commercial areas, industrial areas etc. 1) 2) 3) 4) 5) 6) Rx Level. Rx Quality Timing Advance Handover parameters Data of six best neighbour cells. Layer 2 and layer 3 messages.

From the data collected various information can be extracted which depict the performance of BTS sites and the network as a whole. Following information can be extracted from this drive test data. 1) 2) 3) 4) 5) 6) Coverage edge probability Coverage area probability Speech quality Frequency and BSIC reuse Neighbour cell definition details. Handover details.

Edge Probability:
To get an idea of coverage area, coverage boundary of all the cells based on received signal level (RXLEV), is obtained and is plotted over the geographical map of the area. The coverage boundary of a cell is considered to be made up of equal received level points on the field. The procedure of obtaining edge probability is discussed in ========

With the help of this coverage plot the edge probability or the probability of getting a signal level better than a specific value over the boundary of all cells is obtained which helps in determining the performance of the network with respect to coverage boundary requirements given by the customer.

Area Probability:
The obtained signal levels from the cells at all the points of the network, are then used to make, a best server plot. This best server plot is drawn by categorising it on the basis of in building coverage, in car coverage and on street coverage. These categories are defined on the basis of the coverage area where a good quality conversation is required. The details about threshold defined for these categories are discussed in coverage planning report.

Speech Quality:
Speech quality is a very important aspect for determining the quality of service for whole of the network. Speech quality is inferred by the RXQUAL measurements during the drive test. RXQUAL, is the Bit error rate (BER) derived from the 26 bits midamble on TDMA burst. Its level characterises speech quality where 0 indicates the highest quality and 7 the worst. Thus during drive test, poor quality areas can be found and marked by looking over the quality on the scale of 0 to 7. RXQUAL can be poor due to poor RXLEV, Co-channel interference, adjacent channel interference or multipath. RXQUAL is measured and tested for all the categories of clutter and terrain.

Frequency and BSIC reuse:

From the collected data the frequency reuse pattern with the BSIC (Base Station Identity Codes) planning of all the cells of the network can be obtained. The reuse distance for all the reused frequencies can be determined

Neighbour cells definition details:

With the help of collected data 6 best serving neighbours of all the cells can be determined. The drive test window of the antenna system gives details of 6 best neighbours at an instance.

Handover details:
There are certain other very important parameters which has to be checked during drive test as these parameters directly reflects on the performance of the network, like handover margin, handover threshold, values of handover timers, offset and penalty for the handovers. With a call established, and measuring on the cell edge, we can display the phone measurements of serving and neighbour cells. The difference between the RXLEV of the server and that of neighbours can be monitored on the amplitude and time scale. At

some point on the drive-test route, one of the neighbours RXLEV will become stronger than the servers signal level and when this difference of the two exceeds the handover margin, for atleast a timing set in the handover required counter in BSS, a handover will occur. Thus by simultaneously monitoring RXQUAL during the handover, the value of the handover margin can be determined and a decision can be made whether that value is appropriate for the quality of service desired.

Analysis of extracted data:

The information extracted from the collected data is then analysed to compare it with the agreed benchmarks related to coverage, quality, handover success rate etc and is used to infer the cause of the deviation from given requirements and set benchmarks. It is also used to infer cause of detected problem in the network if there is any. There are special coverage requirement which are discussed in coverage planning report under special coverage category these specific coverage requirements are matched to find out whether the requirement of customer is taken care of or not. RXQUAL is also matched with the given requirement. If RXQUAL is poor and RXLEV is sufficiently good it can reasonably be deducted that the cause is interference. Generally a test frequency which has no adjacent or Co channel present in that area is used to find out if interference is because of multipath. If it is not because of multipath then spectrum analyser can be used to find out whether it is an adjacent channel interference or it can be deducted that it is Co channel interference. A handover margin on the high side will result in a handover occurring after the user has experienced some deterioration in quality. High handover margins can result in poor reception and dropped calls, while very low values of handover margin can produce Ping-Pong effects as a mobile switches too often between cells. With the help of collected data it can be found out weather uplink and downlink are balanced or not. If even after having good RXLEV and RXQUAL, calls are dropping or even when RXLEV and RXQUAL of serving cell is better than that of neighbour cell, handover is taking place, it indicates that the link needs to be balanced. BSIC for all the cells are also checked and verified with what is defined in the BSS. If same BSIC is defined for cells having same BCCH frequency and these cells coexist in the neighbour list then understandably lot of handovers will be unsuccessful. Layer 2 and 3 messages can be used for analysing cause of a particular hadover failure, call drop, very poor speech quality or any other abnormality in the performance of the network.

Suggesting changes in the network configurations based on the analysis:

After detection of the causes of the deviation from the requirement or network related problems, measures are taken to improve the performance of the network and to match customers requirement. Network performance can be influenced by the network parameters. The configuration parameters can be divided into two groups hard configuration and soft configuration, depending on the type of control and action required to modify them.

Hard Configuration:
The hard configuration parameters are aspects of base station configuration and include antenna type, antenna gain, antenna orientation, effective height of antenna radiation centre, use of space diversity, antenna feeder loss and effective isotropic radiated power (EIRP). Changes in this configuration are made to meet the requirements and to deal with the analysed problems. For an example if certain area is affected by interference resulting in poor quality then one of the way to reduce interference level is by shrinking the coverage area. Shrinking of coverage area can be achieved by reducing EIRP that is by replacing the existing antenna with a lower gain or narrower horizontal beam width antenna system and by reducing transmitted power under limitation of not loosing the link balance. Most effective solution used to shrink coverage area is by increasing antenna downtilt and/or reducing antenna height. Similarly to improve coverage in certain areas the transmitted power of BTS can be changed, antennas with different gain or beamwidth can be used and the height of antenna system can be changed. For further specific coverage and quality requirements pico or micro cells can be installed inside the residential places, commercial buildings, stadiums and car parks etc. A pico cell is nothing but a cell with very low EIRP in comparison to a Macro Cell. Note that the neighbour list for these pico cells is defined differently than that for normal Macro cells. Micro cell has also got lesser coverage area than that of Macro cells. Repeaters can also be used for providing coverage to specific areas. There can be Channel selective or Band selective Repeaters where band selective repeaters amplifies the whole GSM band and transmit it towards the area required to be covered while channel selective repeaters receive power from selected channels of one or more than one parent cells, amplify it and direct it towards the area required to be covered. In the similar way if capacity requirement of certain area is more, then the coverage of a cell is needed to be compressed by any of the means discussed above so that it may cater to lesser number of customers.

If the mentioned measures dont work for matching coverage and capacity requirement then relocation or addition of site can also be suggested. If interference is observed during drive test then apart from reducing coverage area, frequency plan for the network can be redefined and reuse distances can be increased. After carefully studying the statistical data about the network performance if it is found that congestion for some particular sites are more and call successful rate is less, then more resources (TRX) can be added to improve availability of the traffic channels or additional BTS sites can also be added but this addition has a limit because of limited available frequency spectrum hence with higher number of sites or frequency used, reuse distance of the sites will reduce which will increase interference and hence the quality will go poorer. There are lots of other ways by which capacity can be increased without much affecting the speech quality. a) Addition of Micro and pico cells. b) Using Underlay and overlay cells. c) Deploying frequency hopping (refer to ---Everytime TRXs are added in the network, frequency plan of the network or a portion of the network has to be changed which will further require to analyse the network using drive test system, to monitor the networks performance. It is possible that after addition of certain TRXs frequency reuse distance will decrease to such a level that it will introduce unacceptable amount of interference and deployed frequency plan will require to be redefined.

Soft Configurations:
Other parts of the system can be controlled with soft parameters. These affect operation of algorithms within the system, and include categories such as common BTS parameters, cell access parameters etc. GSM defines around 150 soft parameters. For an example if it is found from the BSS statistics details that excessive handovers hence more utilisation of resources is taking place then reduction of overlap of the cell coverage areas can avoid them. Defined BSICs for the cells specially for cells transmitting same frequencies are set to be different otherwise lot of unsuccessful handovers will take place. Even then, if it is found that number of unsuccessful handover is high then redefining the neighbour list in BSS can control it. Several neighbours for a serving cell can be defined in GSM. Usually, we want a handover to be made to the strongest neighbour, but in some cases frequent handovers to this best neighbour can result in congestion in this cell, affecting the users initiating calls from that cell. The situation can also occur in reverse, when a handover required to the best neighbour can result in a rejection due to unavailability

of resources, causing the handover to be attempted to the next best neighbour, which can delay the process and deteriorate the quality further. Under certain circumstances, we may need to remove a potential neighbour from the neighbour list and provide alternatives. In the idle mode, the mobile always prefers to remain with or move to the best serving cell. The best cell is decided on the basis of uplink and downlink path balance in the cells. This balance is calculated by GSM defined C1 calculations. C1 calculations force the mobile to move to the strongest cell. In certain cases, such as macro-micro cell architecture, optimisation may require that in certain areas the mobile not remain in the best cell, but instead remain in a cell depending on traffic loading. C2 parameters provide the option of adding fixed positive or negative offsets to the C1 calculation in each cell. So, although C1 might be better for a neighbour cell, the application of C2 parameters could delay reselection. C2 parameters also allow the mobile to apply temporary offsets for a period known as penalty time, which helps reduce Ping-Pong effects. With the help of carefully done drive test these parameters like offset or penalty time for handovers can also be checked and verified. Optimisation of the network using drive test system is an iterative process thus after deploying discussed changes in the network drive test is done again and mentioned steps are repeated until required performance objective is fulfilled. Drive test is very effective part of the optimisation of the network but drive test data is not very effective to find out some of the very specific problems their cause and solutions to rectify them. For finding requirement of capacity, exact cause of handover failures and reduced call success rate one has got to be dependent on statistics obtained from the BSS. The procedure to optimise network using data from BSS statistics is discussed in ----. On the other hand drive test is the only medium with the help of which users perspective of quality of service can be visualised hence simultaneous monitoring of the BSS statistics and drive test data gives most practical and optimum cause and the solution of a problem.