PERFORMANCE EVALUATION AND OPTIMIZATION OF CDMA2000 1X MOBILE CELLULAR RADIO NET WORK 1.

0 Introduction Cellular network operators must periodically evaluate and optimize their ne tworks to accommodate traffic growth and performance degradation. Evaluatio n and optimization action after service rollout is to correct the expec ted errors in network planning and the benefits like improved network capacity, enhanced coverage and quality of service.[1] Wireless Access optimization is the major challenge to every wireless communicat ion system. The classical cellular architecture provides only rough guidel ines, and propagation modeling and statistics, and network simulations fa ll short of representing and optimizing the actual network performance[ 2]. Cost effective wireless access optimization governs the radio coverage at the base stations and additional RF access points as required by the environmental and traffic demands. Optimal performance is achieved b y combining prediction models with dynamic measurements and applying cove rage and diversity means accordingly. Some research efforts in the pas t had tried to show that the optimization of the antennae parameters would be able to yield some capacity improvements, however poor han dling of this operation or insufficient tilt will result in high leve l of inter-cell interference thereby reducing the capacity of the netw ork, and ultimately the network operators could end up paying for unn ecessary network infrastructure[3]. Some previous works are reported on the combined effects of radio interface para meter and network performance indicator at real data [1]-[7]. Some work investig ates individual as well as combined effects of antenna height, high gain antenna s and antenna down tilting for microcellular applications [1]. Using simulations , the authors show in [2] the smaller the cell sizes the larger the antenna down tilt should be, and the higher the traffic load the higher the antenna down-til t should be to improve performance. While the inter-cell interference is indeed reduced through such down tilt, the sectorization efficiency may also be reduced with increased down tilt [3]. Another study finds that optimization of pilot po wers in a CDMA-1X system does not increase the performance of the system signifi cantly and substantial reduction in capacity can be associated with pilot power optimization. However the optimization of the antenna parameters is able to yiel d substantial capacity improvements [4]. Recently [5] has discussed antenna down tilt concepts and performance in an actual operating cellular system, relations hip between antenna height, down tilt angle, and coverage radius. In modern urb an propagation environments, the cell coverage area is less than 2 km it conclud ed. The tradeoff between the coverage and number of users with fixed transmit power is presented in [6]. More recent work, has examined the impact of interference, such as, the problem of where to locate base stations so as to maximize user capacity in a cellular C DMA network is studied in [7]. The impact of intra-cell and average inter-cell interference using power-control constraints is captured. All these efforts had not actually resulted into the desired capacity and covera ge enhancements due to these obvious limitations. i. RF coverage on the forward link is much larger than that in the reverse link because excessive BTS power is allocated to the remote user. ii. Excessive soft handoff area caused by improper cell site layout and misu se of base station antennas. iii. Improper RF parameter settings, as this needs to be fine tuned according to the traffic loading distribution. However this research effort attempts to bridge these limitations through the fo llowing contributions. i. Link imbalance can be identified by drive testing into problematic areas or analyzing the network performance data for problematic clusters/sectors.

ii.

Unnecessary soft handoff could be reduced through the following methods;

Maiduguri network of Visafone communications limited, a privatel y owned and nationally licensed telecommunications operator that started commercial operations in Nigeria in February 2008 using the cdma tec hnology platform is the network under study. Maiduguri, which is the c apital city of Borno state, situated in the North East geopolitical zo ne of the country, with a beautiful flat terrain, is one of the lat est cities Visafone has rolled out services with sixteen BTS (Base sta tion transceiver station) sites. The BTS sites are connected to the Kano BSC (Base station controller) via 16E1 leased fibre link. The scope of this research work includes checking, uploading and analysis of transm ission alarms, RSSI, FER, and other traffic statistics data from the M2000 and the LMT from all the operational BTS in Maiduguri, for acq uisition of initial knowledge of network quality, and to eliminate any existing faults in transmission and installation. Based on this , driv e tests was carried out and analysis of the traffic statistics was m ade, and the network optimization was then implemented at a higher le vel using the various techniques. 2.0 Evaluation and Optimization Approaches Optimization process begins with monitoring, collection of data, and the perfor mance of maintenance functions on the BTS at the far-end, using the Local Mainte nance Terminal (LMT) which is connected to the BSC BAM, known as a client and Back LMT Administrator Module (BAM) known as a server. The BTS Audit is also required to see the real condition and quality of the whole of the device - the device BTS by reference the actual parameter values, because sometimes the value s obtained from monitoring (far-end) is far different from the direct measuremen ts performed on the field. The drive test data collection requires post-process ing of data analysis, and finally the tuning of the network is needed to solve p roblems of RF coverage. Drive test is performed again to validate the earlier ac tions taken. 2.1 Drive Test Exercise To drive test, the following tools could be used; The Agilent E6474A, drive test equipment such as a laptop unit that has installed software for the Agilent CDM A, GPS, two fruit headset (for the long call and short call), power supply, Agil ent hardware devices, and street map site.

Drive test procedure itself was divided into three levels namely Single cell fun ction test, a cluster optimization, and system optimization. Single-cell function tests: Done to test individual base stations. Cluster Optimization: Done for testing multiple base stations in a single cluste r and the performance of good relationship between the BTS. System Optimization: Done to test a wider network performance, e.g. in some clus ters. Drive Test are done on several conditions: Initial test drives conducted when an installed base station have been completed to determine the initial data of a BTS and also to demonstrate the feasibility of a network level. Maintenance drive test conducted in order to monitor the performance of the BTS in accordance with the schedule. Executed for these necessary reasons, i.e. if there were complaints from custome

rs or impairment of performance of base stations as can be seen from the daily r eports. 2.2 Analyzing drive test data In some places drive test data uses Actix analyzer software. Actix analyzer soft ware is useful for processing drive test data, visualize data, analyze the data and can identify network problems for subsequent resolution.

Here are some parameters that are usually seen in the analysis of drive test dat a as displayed by the software in figure 2 RSSI (Receive signal strength indication) Shows the power level received by the MS to the BTS transmitter power in certain areas Ec/Io The ratio between energy produced from each pilot with a total energy received. Ec/Io also shows the minimum power level (threshold) where the MS can still mak e a call. Usually the values of Ec/Io determine when the MS should perform hand off. Tx Power Additions to the value of MS transmit power will cause interference to other use rs. FFER (Forward Frame Error Rate) FFER defined as the average frame error of each frame transmitted. Where the av erage frame error is not more than 2%. Active PN BTS PN is active when doing a drive test. Active PN is defined as an antenna sit e in a state of on-air. From the data obtained and shown again using actix software analyzer can be seen on the RF coverage issues so that later can be searched in the way of handling the right solution to solve network problems. 3.0 Research Methodology 3.1 Research Concept This research work is categorized as a network planning, optimization a nd evaluation effort whose focus is to evaluate the performance of a mobile cellular radio network in terms of coverage, capacity and qual ity, before and after the optimization exercise. This would have been preceded by the network planning stage where the various models are e xposed. Maiduguri city network area of Visafone Communications limited, t hat provides telecommunications services using the cdma2000 1x technology is used as the test bed and is shown in Figure 3

The key radio parameters that determines network performance, namely Tx lev el (transmit power), Rx level (receive power), Ec/Io of the primary pilot, FFER values, and RSSI values are collated and evaluated The Ec/Io o f the primary pilot and the Rx level are used to define the forward coverage range of the system. The transmit power of the MS is used to measure the reverse coverage range. The mean FFER of the forward link is used to determine the quality of the traffic channel. The k ey performance indicators namely, call drop ratio, call setup success ratio, call failure rate, call connection rate, that have direct impac t on the subscribers on the network were collated and evaluated using the industrial best practice values as benchmark for the exercise.

3.2 Experimental Test Bed Following the need for the coverage of Maiduguri city in Nigeria, 16 BTS sites w ere built by Visafone, with the intention of creating a good CDMA 1x network in the city. The BTS configurations are S1/1/1; with frequency point of 419 in the 800MHz band. The 16 BTS sites are connected to the Kano BSC, which is one of the nine BSC locations of Visafone, via a 16E1 capacity fibre link. The list of the on-air sites, their co-ordinates and configurations are listed in Table 1 Table 1: List of On-Air Sites BTS Name Longitude Latitude BTS Configuration MAD001 13.1572 11.8606 S1/1/1 MAD002 13.16081 11.84746 S1/1/1 MAD003 13.14328 11.8492 S1/1/1 MAD005 13.12693 11.8593 S1/1/1 MAD006 13.16807 11.8358 S1/1/1 MAD007 13.14624 11.8242 S1/1/1 MAD008 13.13578 11.8128 S1/1/1 MAD009 13.14646 11.8029 S1/1/1 MAD010 13.15882 11.8149 S1/1/1 MAD011 13.17486 11.81326 S/1/1/1 MAD012 13.19302 11.8088 S1/1/1 MAD013 13.17873 11.8611 S1/1/1 MAD014 13.08002 11.8354 S1/1/1 MAD015 13.09693 11.8253 S1/1/1 MAD016 13.11675 11.8155 S1/1/1 MAD017 13.18319 11.8419 S1/1/1

3.3 Sampling Technique Recently, new sites were installed in Maiduguri. After installation, It was deci ded that DT (Drive Test) should be carried out to determine the performance of t he sites and to proffer solution should any coverage problem be discovered in th e network. The Drive Test was carried out as decided, but certain areas of the network prov ed to have poor network coverage. Hence, optimization was suggested to improve t he network, with antenna adjustment plan as shown in the Table 2 Table2 Antenna Adjustment Proposal

Note: Black stands for values BEFORE OPTIMIZATION Red stands for values AFTER OPTIMIZATION. 3.4 Research Method and Test Route Install a set of drive test equipment with GPS and Qualcomm 1X test mobile stati on into the test car. Besides, a compatible computer installed with collection s oftware (Pilot Panorama). Set the two Mobile stations to the status of Markov lo ng call and short call respectively at full rate. MS is used to test Transmit Le vel (Tx Level), Receive level (Rx Level) and Down link Pilot (Ec/Io) of the prim ary pilot. The Ec/Io of the primary pilot and Rx level are used to define the fo rward coverage range of the system. The transmit power of the MS is used to meas ure the reverse coverage range. The mean FFER of forward link which is used to d etermine the quality of the traffic channel has also been determined. Factual dr ive test route is as shown in figure 5

3.5 Procedure For Data Collection The researcher were part of the 6 man team of Engineers (2 from Visafone, 1 from Huawei Technologies) that carried out the pre-optimization and post-optimizatio

n drive test in the city of Maiduguri, and the following procedures were adopted for the collection of data for this research, Voice long call origination of ou tdoor coverage drive test and Voice short call- origination of outdoor coverage drive test. This is shown in Table 3 and Table 4 respectively. The RSSI data wer e pulled from the M2000 server i. Voice Long Call Origination Of Outdoor Coverage Drive Test Table 3 Case of Long Call Prerequisites MS is ok. Test laptop with Dingli software installed Select the test routes Duration 300min Action Connect the ms and GPS to the computer, and make sure its ok. Start Dingli application software. Start logging at the mobile station Reception the call automatically. Highlight the Original call Service option: Markov Call setup timer: 10s Conversation timer: 99999s Call tears down timer: 5s Pause timer: 5s Redial while drop: Yes (checked) Start driving test follows the selected route 300 minutes later, end the call. End logging at the mobile terminal, and record the mobile log. Finish the Voice Long call-Reception of Outdoor Coverage Drive Test Cleanup Release the call

ii.

Voice Short Call Origination of Outdoor Coverage Drive Test Table 4 Case of Short Call Prerequisites MS is ok. Test laptop with Dingli software installed Select the test routes Duration 300min Action Connect the ms and GPS to the computer, and make sure its ok. Start Dingli application software. Start logging at the mobile station Reception the call automatically. Highlight the Original call Service option: EVRC Call setup timer: 10s Conversation timer: 30s Call tears down timer: 5s Pause timer: 5s Redial while drop: Yes (checked) Start driving test follows the selected route 300 minutes later, end the call. End logging at the mobile terminal, and record the mobile log. Finish the Voice Short call-Reception of Outdoor Coverage Drive Test Cleanup Release the call 4.0 Results and Data Analysis The data were collated through the drive test, before and after the network opt

imization exercise .These were processed with the drive test tool, Dingli Panora ma and the output was produced in the form of distribution charts and maps and s ubsequently analyzed to lay credence to the set objectives of the study. This re sult was compared with the simulation output produced by the simulation software , Tems cell Planner. These outputs of the examined network optimization metrics include, Ec/Io, Rx level, Tx level, FFER, Call drop, PN Plot, Short call statist ics. 4.1 Ec/Io Coverage The diagrams in Figure 6, 7, 8, and 9 show coverage by Ec/Io in Maiduguri city b efore and after optimization was carried out in the network. Areas around MAD016 and MAD 002 had poor coverage due to GPS antenna tracking problem of MAD0016 an d MAD002 coupled with some sites with RSSI problems. However, after optimization , all these snags were cleared and the city has excellent Ec/Io coverage. As sho wn by statistics in Figure 10, Total Ec/Io greater than -9dB was 91.51%, before optimization. However, after optimization, as shown in Figure 11, the value came to 95.08%, which is a remarkable improvement over the previous value for the ne twork.

Fig 9 Max Ec/Io distribution map after optimization

5.2

Rx Level Coverage

Areas around MAD 016, MAD 002 and MAD011 had poor coverage. But after optimizati on, as shown in Figure 13, Rx Level has improved in the affected areas. Figure 14 shows that Rx level above -90dBm was 99.21%, before optimization. Howe ver, after optimization Rx level above -90dBm came to 99.41%, which is an improv ement over the previous values as shown in figure 15

5.3 Tx Level Coverage As shown in Figure 16, the Tx distribution in most areas is good except for the area around MAD 016, which is at the edge of the coverage. However, after optimi zation, the Tx improved significantly in this area as shown in Figure 17. Figure 18 shows that Tx level below 0 was 93.52% before optimization. However, as show n in figure 19, after optimization Tx level below 0 came to 96.14, which is a go od improvement over the previous value

Fig 17: Tx Level distribution map after optimization

5.4 FFER Coverage As shown in Figure 20, the quality of the network measured by FFER was poor in a reas around MAD 016 and MAD 002. However, as shown in Figure21, this improved af ter optimization .Figure22 and Figure 23 reveal that the value of FFER value low er than 2% was 81.51%. But after optimization, the value came to 83.7%, which is a good improvement over the initial value.

5.5

Call Drop (Long Call)

As shown in Figure 24, before optimization, call drops featured at the center of the city, as well as at areas toward the edges of the network, GPS tracking problem on some sites. However, as shown in Figure 25, no call dro ps featured in these areas and the rest of the network after optimization.

5.6

PN Plot Before optimization, there was an antenna swap between sector 0 a

nd sector 1 of MAD012 as shown in Figure 26. However, the antenna swap was corre cted. From Figure 27, it can be seen that after optimization, the serving cells maintain their correct PN values, including the area situated at the edge.

5.7

Short Call Statistics Table 5 Short call statistics before optimization Report Message Count Percent Call Success 414 98.57% Call Drop 6 1.43% Call Connection 426 96.60% Call Failure 15 3.40%

Table 6 Short Call Statistics after optimization Report Message Count Percent Call Success 559 99.82% Call Drop 1 0.18% Call Connection 576 99.13% Call Failure 5 0.87% As shown in the Table 5 and Table 6 there is an improvement after the optimizat ion in the Call Setup success ratio and call drops. 5.8.1 Analysis of Simulation output Downlink Received Power Figure 28 shows the total mean received RF power over a carriers bandwidth, found by adding the received pilot powers, common channel powers, traffic channel pow ers and background noise at any pixel. The value at each pixel is obtained by av eraging the total received powers of the terminals attempted at a pixel. In the plot, different colours are applied to identify the different strength level of the Mean Received Power. For example, the Green represents the Received Power le vel between -80dBm and -70dBm.In CDMA system, downlink coverage effect lies on b oth DL Received Power and Pilot Ec/Io. For example, with the increase of user, t he strength of Received Power will be stronger, while the strength of Pilot Ec/I o will decrease. Only if both of them satisfy the coverage threshold, the downli nk coverage effect could be ensured well. The outdoor coverage threshold of Rece ived Power is -105dBm; meanwhile in this simulation, there is 5dB-fading margin to obligate for 75% edge coverage probability, which means in the area where the Received Power is greater than -100dBm the outdoor edge coverage of downlink ca n be satisfied with a probability of 75%. Moreover, here we set the penetration loss of buildings as 20dB, so the indoor coverage threshold of Received power is -80dBm. As shown in the plot, the Received Power in the areas except orange, ye llow is greater than -80dBm, so the downlink indoor coverage in these areas can be satisfied. [21]

The Uplink (UL) Required MS Transmit Power Figure 29 shows the mean uplink required Tx Power at each pixel for the selected service. In the plot, different colours are applied to identify the different s trength level of UL Mobile Transmit Power. For example, the green represents the UL Transmit Power level between -22dBm and -12dBm. The outdoor coverage thresho ld of the UL Mobile Transmit Power is 23dBm; meanwhile in this simulation there is 5dB-fading margin to obligate for 75% edge coverage probability that means in the area where the UL Mobile Transmit Power is less than 18dBm the uplink outdo or edge coverage can be satisfied with a probability of 75%. Moreover, here we s et the penetration loss of buildings as 20dB, so the indoor coverage threshold o f UL Transmit Power is -2dBm. As shown in the plot, the UL Transmit Power in the red, orange, yellow, green, blue area is less than 2dBm, so the uplink indoor coverage in these areas can be satisfied. [21]

The Downlink (DL) Pilot Ec/Io The Pilot Ec/Io figure displays the achieved pilot Ec/Io for each pixel and is i n effect, the same as Ec/Io for the first best server. In the plot, different co lours are applied to identify the different level of DL Pilot Ec/Io. For example , the Green represents the Pilot Ec/Io level between -9dB and -7dB. In CDMA system, the Tdrop threshold is 13dB, which is to say that when the Pilot Ec/Io level is below this threshold, the DL coverage cant be satisfied; and when the DL Pilot Ec/Io level is below 11dB, the DL coverage effect may be not good. A s shown in the plot, the Pilot Ec/Io in the pink, blue, green, yellow, orange ar ea is greater than 13dB, so the downlink outdoor coverage in these areas can be s atisfied; and the pilot coverage effect in the pink, blue, green area, yellow is good.

6.0 Conclusion The optimization process is a long term process that requires the study of the n etwork situation and the provision of solutions to weak features sorted out firs t, without a hasty implementation, for a successful outcome. The antenna hardwar e changes (tilt and azimuth) are important issue in the network optimization, as it is observed that most times the advised changes are not correct. The load on the system increases with time and thus affecting the network performance, henc e the need to periodically monitor the carrier loads, and expand the network if necessary. Interference affects network capacity and the overall performance and quality of end user experience (call setup, call drop rate, etc) and these are considered key issues that need to be resolved. Network planning must be based o n standard value to predict the demand services area and Qos. It is important to know the network layout and Qos before implementing optimization procedure. The re are radio elements to use as check to adjust the parameters to enhance the Qo s. The threshold values of the parameters must be used as performance indicators to effect radio interface optimization. Knowing the radio interface protocols o f cdma2000 1x is essential for radio interface optimization. The function of cha nnels in setup call and the messaging can provide one with the reasons behind hi gher call setup failure and also high call drop rate. REFERENCES:

[1] Edward B., and Abu S.,(2004), Effects of Antenna Height, Antenna Gain and Pattern Downtilting for Cellular Mobile Radio, Proceedings of IEEE infocom 2004. [2] George C., and Matt D.,(2006), Antenna tilt control in CDMA Networks, Moto rola Inc. 2006. [3] Scott B.,(2003), Antenna Downtilt Effects on CDMA cell-site Capacity, US W est Advanced Technologies, 4001 Discovery Boulder, CO 80303. [4] Chris M., Matt D., Aniruddha D., (2001), Performance Gains Using Remote C ontrol Antennas in CDMA and 1xEV-DO Networks,Motorola Ltd, Motorola Inc, Motorola India Electronics Ltd. [5] Wu J., Yuan D., (2005), Antenna Downtilt Performance in Urban Environment s, Mobile Communication branch, Jinan Telecomm. Bureau, Jinan, Shandong, China. [6] Venugopal V, and Andrew S, (1999), The Coverage-Capacity Tradeoff in Cell ular CDMA Systems, IEEE Tran 1999. [7] 3GPP2 C.S0002-B, (2002), Physical Layer Standard for cdma2000 Spread Spec trum Systems, v1.0, April 2002. [8] 3GPP2 C.S0002-C, (2002), Physical Layer Standard for cdma2000 Spread Spec trum Systems, v1.0, May 2002. [9] 3GPP2 C.S0024-C, (2002), cdma2000 High Rate Packet Data Air Interface Spe cification, v4.0, October 2002. [10] 3GPP2 C.S0010-B, (2002),Recommended Minimum Performance Standards for cdm a2000 Spread Spectrum Base Stations, v4.0, December 2002. [11] 3GPP2 C.S0011-B, (2002), Recommended Minimum Performance Standards for cd ma2000 Spread Spectrum Mobile Stations, v4.0, December 2002. [12] <www.tektronix.com/signaling_test, (2003), CDMA Network Technologies: A D ecade of Advances and Challenges, June 2003. [13] Esmael D.,and Aleksey K., (2006), The Impacts of Antenna Azimuth and Tilt Installation Accuracy on UMTS Network Performance, Bechtel Corporation, January 2006. [14] Huawei Technologies Co, Ltd, Red Treasured Book for Onsite Optimization o f CIS CDMA June 2005 [15] Stefan K, (1998), Multi-Carrier CDMA Mobile Radio Systems Analysis and Opt imization of Detection, Decoding, and Channel Estimation, German Aerospace Centre (DLR) Institute of communications and Navigation, January 1998 [16] Rosemary B, Shalinee K, (2005), Network Upgrade Design in Tiered CDMA Cel lular Networks, June 2005 [17] Scott C. Bundy, (1999), Antenna Downtilt Effects on CDMA Cell- Site Capac ity, US West Advanced Technologies, IEEE Publications 1999. [18] Myron D. Fanton, (2006), Antenna Pattern and Coverage Optimization, PE ERI Technical Series, Vol. 6, April 2006. [19] Joseph Shapira, (2004), Enhancement and Optimal Utilization of CDMA2000 N etworks, CEWIT CDMA Workshop, Nov.2004 [20] Huawei Technologies Co., Ltd, (2003), Network Preliminary Proposal for Er bil CDMA WLL Project, 2003 [21] Coperation, (2006), The Wireless Network Planning Report For Dalyasat Pro ject, 2006. [22] HuaweiTechnologies Course Development Room, (2003), RG0105 CDMA 1x Air In terface, March 2003 [23] Nortel Network, (2001), CDMA Technology Overview, February 2001 [24] ZTE Corporation, (2006), Technical Specification of ZXC10 BSCB, Shenzhen P .R. China, 2006. [25] Dingli Communications Inc, (2004), Pilot Panorama User Guide, August 2004. [26] ZTE Cooperation, (2006), CDMA Principle, November 2006 [27] Achim W., Kari S., Antti K., (2001), Auyomatish and Remote Optimization o f Antenna Subsystem Based on Radio Network Performance, Nokia Networks, 2001. [28] 3GPP2 C.S0009-0, (1998), Speech Service Option Standard for Wideband Spre ad Spectrum Systems, TIA Publication, 1998. [29] Mel Menke, (2003), Challenges Facing CDMA Network Optimization, CDMA Servi ce Product Manager, Motorola Global Telecom Solutions Sector, 2003 [30] John P., and Asim Q., (2004), IS-95 CDMA Forward Link Optimization Tool, A ir Touch Cellular, 2004

[31] 3GPP2 C.S0001-D, (2005), Introduction To CDMA2000 Standards for Spread Sp ectrum Systems, September 6, 2005 [32] 3GPP2 C.S0005-A, (2002), Upper Layer (Layer 3) Signaling Standard for CDM A2000 Spread Spectrum Systems, February 2002 [33] 3GPP2 C.S0003-D, Medium Access Control (MAC) Standard for CDMA2000 Spread Spectrum Systems, September 2005 [34] Dongmei Z., Xuemin S., and Jon M., (2006), QoS Guarantee and Power Distri bution for Soft Handoff Connections in Cellular CDMA Downlinks, IEEE Transactions on Wireless Communications, Vol. 5, No. 4, April 2006 [35] Tansu A., Xingzhe F., Tamer B., Murat A., and Jon T.W, (2004), Power Cont rol for Multicell CDMA Wireless Networks: A Team Optimization Approach, Universit y of IIIinois at Urbana-Champaign, 2004. [36] Robert AKL, (2006), Subscriber Maximization in CDMA Cellular Networks, Dep artment of Computer Science and Engineering, University of North Texas, CCCT04, 2006 [37] Hubaux J., (2007), Call Setup CDMA Mobile Networks, Module D Cellular Net works, 2007. [38] Kathryn Oliver, (2004), Introduction to Automatic Design of Wireless Netw orks, Centre for Intelligent Network Design, Cardiff University, UK, 2004. [39] Michael J Martin, (2006), Rf Physic: the rules are changing, IBM Canada, J anuary 2006. [40] Viterbi A.J., (1995), CDMA: Principles of Spread Spectrum Communication, R eading, Wesley, 1995