the rapid growth of cellular networks in recent years produces the effect to optimization of networks, so that they provide a better service to customer without any interruption. The needs of a Cellular operator include delivery of data and voice properly to the Mobile Stations or user equipments. Data services are gaining more significance in mobile communication systems like UMTS; speech telephony remains the most frequently used service. Speech quality experienced by the end user can therefore be regarded as one of the most significant quality aspects of mobile communication. They have to detect all the problems faced by the mobile network. The main motive of cellular operator is to provide and to communicate a high service level towards the customers.

Keywords-speech quality index(SQI), key performance indicator(KPI), drive test (DT)

Introduction

Monitoring and evaluating in cellular network is fundamental to the success of every cellular operator. Measurements help them to meet their Objectives. Key Performance Indicators (KPIs) is one important tool in the cellular network. KPI gives good indication of network performance. KPIs are organized data collecting and collating, where the Operator can monitor and evaluate their performance and drive for continuous improvements towards meeting their objectives.[1] Whereas Speech Quality Index is a method to measure and monitor the speech quality in the network based on radio quality information.[2] The use of SQI technology for GSM and WCDMA is increasing, and there are compelling reasons to use this technology. SQI is a sophisticated measure which is dedicated to reflecting the quality of the speech. The speech quality on the uplink is determined by continuously monitoring the uplink radio conditions for each ongoing call in the network. The information element SQI (Speech Quality Index), which we obtained estimates the speech quality in the cellular network as perceived by a human listener. In this paper we explain how SQI algorithm works for GSM. We can design the SQI algorithm for WCDMA in the same spirit, and its implementation is similar though necessarily not identical.[3] Speech quality in GSM networks is measured by means of the many parameters like RxQual, BER, FER and L3 messages. Traditionally we use RxQual parameter to measure the speech quality, however, suffers from a number of drawbacks which make it an unreliable indicator of speech quality.[4]

There are two different measurements of speech quality in communication networks performed with intrusive and non-intrusive means. The proposal made in this paper is described SQI measure, speech quality measurement methods that are categorized to highlight differences between measures, performance evaluation figures presented, and finally a motivation why speech quality is possible to measure in the given cellular network.[6]

Intrusive method:

This method to measure speech quality is based on a comparison of both, original and transmitted signal using proper algorithm, e.g. ITU-T P862 (PESQ). By using intrusive method it gives more accurate results in comparison with the quality assessed by the average listener that is acquired from listening tests. Due to high cost and time consumption forms a space for the second, Non-intrusive method.

Non-intrusive method:

The non-intrusive method estimates the quality just from the transmitted sample.[7] It is easy to see the difficulty of such an algorithm to give reliable results. Research into non-intrusive quality measurement usually entails comparisons between the test signal and normative behavior of clean speech, amongst other distortion-sensitive features.[8] The main advantage of the non-intrusive method is cost efficiency; the measurement is conducted within real network data and states. We can measure “unlimited” number of speech samples that can be assessed for the quality for the total accuracy improvement. There is no need to call establishment unlike in the intrusive method. It is therefore recommended to use a combination of both, intrusive and non-intrusive methods. Vocal quality is an interaction between an acoustic signal and a listener.

NETWORK PERFORMANCE

Every cellular network provider uses their Performance measurement that is fairly unique. There are many parameters that can be measured. These measurements can be correlated to each other, so the number of permutations is infinite.[11] And there for the question is arise that which is the right choice. KPI are defined by the network operator based on their business needs. And these network optimization targets KPI which are derived from the behavior of the network. Services will be optimized and detected errors will be eliminated. All in all it is correct to say that the purpose of performance measurement is to troubleshoot and optimize the network. Optimization involves monitoring, verifying and improving the performance of the radio network. Its fitfully from the last phase of radio network planning, i.e. during parameter planning. [12]. A cellular network covers a large area and provides capacity to many people, so there are lots of parameters involved that are variable and have to be continuously monitored and corrected. As network is always growing through increasing subscriber’s numbers and increases in traffic. This means that the optimization process should be on going, to increase the efficiency of the network leading to revenue generation from the network. Radio Network planner’s first focus on three main areas

Coverage

Capacity

Frequency planning

After that they have to focus on

Site selection

Parameter planning etc.

In the optimization process there are many issues arise which are addressed with differences that sites are already selected and antenna locations are fixed, but subscribers are as mobile as ever with continuous growth taking place.As the majority passes Optimization task becomes more and more difficult.Once a radio network is designed and operational, its performance is monitored.[13] we choose the key performance indicators to compare the performance. The main focus of radio network optimisation is on areas such as

Quality control

Power Control

Subscribe traffic

Handovers

Resource availability measurements

KPI is defined as key performance indicator for NP(Network Performance) or DP (device performance)[14]. In simple words we can say that KPI is value which is computed using formula. KPI is always a significant assessment standard for network performance for wireless communication. There can be different kind of input for the KPI. There is different kind of KPI and these KPI can be correlated with each other.. Due to lack definition framework and methods for KPI, it’s difficult for existing network performance KPI to cover all important aspects of network to make fair evaluation of network performance. In Cellular network KPI can be classified in three categories [15] which are:

Accessibility

Retainability

Quality

Accessibility has to do with user being able to access the radio resources which include set up a call or data assignment channels. Retainability covers how good the connection is and the ability to keep up a call. Quality can be measured with speech frame error rate. They can all be measured at the radio level and easily be transformed into a measurement of service quality.

PERFORMANCE INDICATORS

RSSI

Firstly, the most important and arguably consideration in radio link managment is signal strength. In cellular network, received signal strength indicator (RSSI) is a measurement of the power present in a received radio signal. RSSI is radio receiver technology metric, which is usually invisible to the user of the device containing the receiver, but is directly known to users of wireless networking .

RxQual

RxQual is considered as a basic measurment. RxQual is a value that represents the quality of the recived signal. The values are based upon all frames on the SACCH multiframe, where the bits may have been garbled along the RF path or lost due to fading or interference. This means that if Discontinuous transmission (DTX DL) has been used, the values will be invalid for that period since they include bit-error measurements at periods when nothing has been sent resulting in very high BER. It simply reflects the average BER over a period of 0.5s. the number of bit errors is accumulated in a BER sum for each SACCH muli-frame and the result is classified from 0 to 7 according to the BER- RxQual conversion table. We use RxQual in GSM and is a part of the NMR (Network Measurement Reports).

C/I

The carrier-to-interference ratio (C/I) is the quotient between the average received modulated carrier power C and the average received co-channel interference power I, i.e. cross-talk, from other transmitters than the useful signal. In GSM we often use the Rxlev instead of RSSI. The distinction in these term is that we generally use RSSI for anlog signal and Rxlev for digital networks.

Bit error rate (BER)

In cellular network bit errors rate is the number of received bits of a data stream over a communication channel that have been altered due to noise, interference, distortion or bit syncornization errors. The bit error rate or bit error ratio (BER) is the number of bit errors divided by the total number of transferred bits during a studied time interval. BER is a unit less performance measure, often expressed as a percentage.

Handover

Handover success rate gives the percentage of the successful outgoing handover attempts. A higher value of the ‘handover failure’ would result in higher values of the DCR. Droped Call Rate is the ratio between the drop in traffic channels during the conversation to the number of successful ‘seizers’ on the cells or areas.

L3 messages

Layer 3 messages manages all function necessary to circuit switched call control. The functions performed at this layer are call establishment procedures for mobile-originated calls and mobile-terminated calls, in-call modification, call reestablishment, dual tone multi frequency(DTMF) control procedure for DTMF transmission. The contents contain by this are call control (CC), short message service(SMS) and supplementary service(SS). Simple example of L3 message is paging procedure which is shown in fig below

Figure 1: Paging Procedure

Different methods to evaluate measuring speech quality

The MOS Scale

A common measure for subjective speech quality is the Mean Opinion Score (MOS) scale, defined in the ITU-T standard P.800.[16] In MOS test we have a person who listen to short speech sample consist of four to five sentences. On the basis of this test the person grade the sample from number 1 to 5 where 1 is percived as lowest quality and 5 is highest percived quality.

Modulated Noise Reference Unit (MNRU)

Modulated Noise Reference Unit (MNRU) Scale is defined by the ITU-T standard P.810.[17]. MNRU describes how a speech sample can be distorted in a mathematically deterministic way. We add multiplicative band-limited white noise to get distortion. . Extra samples have been distorted by adding different amount of MNRU distortions. MNRU output is combination of input speech and noise.

Perceptual Evaluation of Speech Quality (PESQ)

“Perceptual Evaluation of Speech Quality (PESQ)” implements an algorithm which compares an original speech sample with another recorded speech sample. Original speech sample known as “reference signal” and recorded speech sample known as “degraded signal”. We get the MOS like score in PESQ by identifying the differences, and by modeling the characteristics of the human perception.

Figure 2:

PESQ assumes that time alignment delay of system is piecewise. Delay changes are allowed in soundless periods and in speech. PESQ maps the signals into a representation of perceived loudness in time and frequency. It include various steps like bark spectrum, frequency equalization, equalization of gain variation and loudness mapping. Absolute difference between the degraded and the reference signals gives a standard of audible error. [18]

We can localized errors dominate perception. In PESQ we integrates disturbance over several time-frequency scales using a method designed to take optimal account of the distribution of error in time and amplitude.[19] The disturbance values are aggregated using an Lp norm, which calculates a non-linear average using the following formula:

The disturbance is first summed across frequency using an Lp norm, giving a frame-by-frame measure of perceived distortion. This frame disturbance is multiplied by two weightings. The first weight is inversely proportional to the instantaneous energy of the reference, raised to the power 0.04, giving slightly greater emphasis on sections for which the reference is quieter, the frame disturbance is bounded with an upper limit of 45. The second weight gives reduced emphasis on the start of the signal if the total length is over 16s, modeling the effect of short-term memory in subjective listening. This multiplies the frame disturbance at the start of the signal by a factor decreasing linearly from 1.0 (for files shorter than 16 seconds) to 0.5 (for files longer than 60 seconds).

After weighting, the frame disturbance is averaged in time over split second intervals of 20 frames (approx 320ms, accounting for the overlap of frames) using Lp norms. These intervals overlap 50%, and no window function is used. The split second disturbance values are finally averaged over the length of the speech files, again using Lp norms. Thus the aggregation process uses three Lp norms – in general with different values of p – to map the disturbance to a single figure. The value of p is higher for averaging over the split second intervals to give greatest weight to localized distortions. The symmetric and asymmetric disturbance is averaged separately.

EVALUATION AND PERFORMACE CRITERIA

The aim of our work is to develop a model for speech

quality estimation, based on parameters available in the Operation and Maintenance Centre (OMC) and in the measurement reports from mobile stations.[23] Such a model will immediately provide an estimate of speech quality for any part of the network, since the measurement reports of the parameters are always available for the whole network.

Usually the following tasks are consider by celluar operator to improve the QoS of network:

1) firstly they have to improve the existing network coverage and capacity.

2) then improve the offered service quality for fulfillment of customer demands.

3) finally maintain the KPIs under pre-defined threshold, and To standardize and benchmark the network performance with that of competitor’s network to attract more customers; keeping a balance between cost and quality.

As the users in the existing network become more and more resources become limited.[25] All the network events such as call initiation, call drop, traffic channel assignment, traffic channel demand, traffic channel release and many more are being reported to the BSC by BTS. Different counters (CT) are being triggered against different events. An event either increments or decrements a specific counter value. The counters are computed by BSC over a measurement period typically one hour and stored in Operation and Maintenance Centre for Radio Network (OMCR database). OMCR is also used to configure and control different Base Station Subsystem (BSS) entities and relevant RF parameters. These counters are collectively used in KPI formulations.[26] Now the Counter data i.e which we collect is in the raw from and doesn’t depict any useful information until and unless it is interpreted using some formulation in the form of a KPI. Hence evaluation criteria use these collected data and KPIs to depict network QoS as a whole. KPIs can be taken as a tool to measure network performance where different KPIs merge together to form a report from this we measure the quality of speech. SQI reports are generated (using these KPIs) on daily, weekly, bi-weekly and monthly basis. Moreover, KPIs are used as a basic unit for network QoS monitoring.

Another method to evaluate the performance of single BTS or Cluster of BTS is through field trials and surveys via Drive Test (DT). During DT, radio network optimization teams use specific equipment supported by Global Positioning System (GPS),special type of handset known as drive test phone having the special funtionality to report command set[27], and laptop with software tool to check the network performance for a specific geographic area. Based on these GSM Network service providers analyze the network performance and evaluate service quality indicators. These indicators can be used for the following mentioned purposes:

1) To identify and locate BSS occasional faults to gurantee physical resource availability.

2) To help to analyze the radio situation, detect radio network problems in one or more BTS and finally devise a way to optimize the network and adopt corrective actions like new frequency allocations, antenna tilt adjustment, and parameter modification in OMCR database etc.

3) To monitor system variance and behavior in terms of

successful attempts, congestion,traffic load etc.

4)To predict the upcoming traffic evolution and network expansions as per increasing number of mobile users.

5) To attract more users at the cost of better Quality benchmark network with another competitor’s network.

To measure and monitor the network performance, BSS vendors have implemented specific software application/feature which is integrated with OMCR and post processes the raw counters into readable information. These software/application can be of two types one is link analysis in which we monitor the live performance of the network and other post analysis in which we record the live data and after that we analyise that data.The purpose of performance evaluation is to analyze all traffic stats and DT results and then an optimization plan is prepared containing all the changes, BTS/BSC parameter modification and physical optimization (if necessary) including antenna mechanical and electrical tilt adjustment etc.

After the preparation of optimization plan, next step is to implement it over a cluster of cells and then measure the Speech quality index using test MS for a specific period of time. Results of changes are then compared to the reference threshold values; if the results are satisfactory then the proposed changes are approved & final implementation is done over that cluster of cells.

CONCLUSION

In this paper, we judge the network performance and evaluate the Quality of Service (QoS) regarding end user perspective, all GSM operators use Key Performance Indicators (KPIs). With the help of these KPIs we derive the different formulations. Various parmeters have been mentioned and in order to evaluate performance,we mainly outlined these parameters applicable to find the formulation to all of them. Using these formuation we can find the speech quality of the cellular network.