ࡱ> %` 0LbjbjNN ,,_4r5r5r5r5d5<}$66666m<o=L=(|||||||$h|Ey@<^m<y@y@|66]=}AAAy@(V66|Ay@|AAt H,y66 Yr5@v6|$S}0}v.@l,y,yz=u>hA>T1?H===||uA===}y@y@y@y@$&Dj'd&j' Implementation of the Software Quality Ranks Method in the Legacy Product Development Environment L. Hribar, A. Burilovi and D. Huljeni Ericsson Nikola Tesla R&D Center, Croatia  HYPERLINK "mailto:lovre.hribar@ericsson.com" lovre.hribar@ericsson.com,  HYPERLINK "mailto:ante.burilovic@ericsson.com" ante.burilovic@ericsson.com,  HYPERLINK "mailto:darko.huljenic@ericsson.com" darko.huljenic@ericsson.com Abstract Software Quality Ranks (SQR) is an important method to manage and improve software quality. Component software quality has a major influence in development project lead-time and cost. SQR enables better management and visibility of the quality effort associated with the component implementation. It also provides a roadmap for continuous improvement leading to value add quality attributes like low maintenance, self-optimizing software and short development lifecycles. SQR method focuses attention to prioritizing the quality investment on design component level through different quality assurance mechanisms (basic test, code review, desk checks, documentation and other actions). The resulting design delivery to verification phase will be more predictable quality software with shorter lead-time and Time-To-Market (TTM). Introduction It is always about quality and cost. Or even better, it is always about ratio between quality and cost. Software quality prediction helps minimize software costs. The number of faults in a large software project has a significant impact on project performances and hence is an input to project planning [1], [2]. The ANSI/IEEE definitions for quality control is measuring the quality of a product and the Quality Assurance measures the quality of processes used to create a quality product. IEEE Standard 12207 defines QA this way: "The quality assurance process is a process for providing adequate assurance that the software products and processes in the product life cycle conform to their specific requirements and adhere to their established plans." [10]. The Software Capability Maturity Model version 1.1 states it this way: "The purpose of Software Quality Assurance is to provide management with appropriate visibility into the process being used by the software project and of the products being built". The ISO/IEC 9126 product quality standard [11] is used to categorize the metrics. ISO 9126 (ISO 1991) is called software product evaluation: quality characteristics and guidelines for their use. In this standard, software quality is defined to be the totality of features and characteristics of a software product that bear on its ability to satisfy stated or implied needs. Software quality is decomposed into six factors: functionality, reliability, efficiency, usability, maintainability, and portability. As the quality level of the final product is set at the beginning of the project, a large number of faults can result in project delays and cost overruns. Implementation of the SQR within AXE defines eight different ranks for different quality levels and a set of aspects within these ranks. Software components can be measured and compared using these aspects. The SQR method presented in this paper are used not only within Ericsson Nikola Tesla R&D center, but also in other Ericsson R&D centers worldwide. Moving components up through to the higher ranks should offer further returns on investment. Minimizing the expenditure for QA yet meeting the customer's quality requirement is not a simple matter [12]. More critical components of the system should aim to have higher ranks. The lower ranks should offer lower fault slip through [3], [8]. While the higher ranks offer more stable and cost effective software component.  SHAPE \* MERGEFORMAT  SQR focus To reach a specific rank, all criteria for that rank and all criteria for any lower ranks have to be met. Criteria are put on every single SQR aspects and are judgment by the experts in every single aspect. For example, a judgment for a structural aspect is evaluated by the experts from that field, while the functionality aspects are evaluated by the experts from that field. It must be noted that criteria is the same for all software components in the organization. The minimum rank is 1 (at the low end) and the maximum rank is 8 (at the high end) and is ranked by the software engineering experts from different experts fields. The quality rank reflects at the low end that the software component is successfully built on simulated environment, no reviews has done, the product is not documented and the quality is estimated. SQR level 1 is usually SQR level for the prototypes. At the high end, it shows that the component is thoroughly tested and designed, and performs according to its specification. On figure 1 is presented focus of the SQR method which is the quality of the design component during project phase of the product development. This article deals with implementation of the SQR method in the Ericsson Nikola Tesla R&D. In section II focus is on the SQR aspects. In section III focus is on definitions for the software component and implementation in the project. Finally, in section IV is conclusion. SQR aspects For each specific quality rank, four aspects have been chosen to describe the quality level of a software component. These aspects are Maintainability, Structure, Functionality and Capability. Maintainability aspect The Maintainability aspect reflects how well known and how well described the software component is. Because the interpretation of and name for the lowest level is imprecise, the term "component" has been chosen rather than other common synonyms such as "unit", "module", or "program" to avoid confusion with these more common terms and remain compatible with them. A software component is a unit of composition with contractually specified interfaces and explicit context dependencies only. A software component can be deployed independently and is subject to composition by third parties [16]. A software component is written in accordance with a specification. It does not matter what the specification is, as long as the component adheres to it. A component specification is a description of a component's function in terms of its output values for specified input values under specified preconditions. Within this aspect it points out by whom the information should be known; the designer, the team or some other group. It also states where it should be documented, commented in the code or as a formal description. It should also state what kind of information you need to document, e.g. the design principles and design patterns. The way to enforce understanding and knowledge of the code is reviews. Depending on the method chosen (walk-through, review, or inspection) and what the review shall focus on, the component will be analyzed differently. IEEE 1028-1997 defines five different types of reviews, Management review, and Technical review, Inspection, Walk-through and Audit [6]. The Inspection and Walk-through apply to component reviews. The difference between the Walk-through and the Inspection is how formal the process is. The Walk-through is conducted with the responsible engineer/author leading the review while the inspections will be led by another assigned and trained leader. The objectives for the Walk-through are to find anomalies, examine alternatives, improve product and to be a forum for learning. The Inspections objectives are slightly different by adding the objective to verify resolution and to verify product quality. Conforming to the standard, the walk-through shall include input of statement of objectives for the walk-through, and the software component being examined. The input to the inspection shall include a statement of objectives for the inspection, the software product to be inspected, documented inspection procedure and inspection reporting forms. It will handle observation on software component, profiling and efficiency measurements and the characteristics of the behavior of the component. The software component code shall contain enough comments to help an experienced designer with no pre-knowledge of the code to understand and be able to do corrections and smaller alterations in the code. A static analysis tool automatically checks the code and provides warnings when it detects deviations from, e.g. coding rules. As fewer and fewer suppressions of warnings are accepted at each higher rank, the maintainability will increase and the failure (actual deviation of the component or system from its expected delivery, service or result) of the software component will be at very low level [9]. Structure aspect The Structure aspect reflects how well the software components are composed within an overall architecture. The aspect also looks at how dependencies communicated and handled across components. The flow of events is the description of how the system reacts and responds to stimulation from the outer world. It gives a vocabulary to describe the decomposition of the functionality into disjoint groups and therefore gives a way of partitioning tests and expectations. Based on the IBM Rational Unified Process [5] terminology for use cases we will use the following terms: A Basic Flow is the normal, expected path through the component. There might be a collection of basic flows that is executed to demonstrate that the functionality works sufficiently. These Basic Flows will not invoke any error handling. An Optional Flow is the normal, expected optional behavior of the system. The system should be able to handle these predictable situations. An Exception Flow is the execution flow in the system to handle unexpected but predictable errors. This will include handling of erroneous data and lack of resources in the environment. The system should be able to handle these situations without problems in the execution. Dependencies across 3rd party interfaces need to be well understood and communicated as well as quality assured to the same level as the dependant components. Dependencies covering detailed explained functionality shall be documented. All dependencies and impacts on other parts of the system shall be inspected by independent team (e.g. inspection body) and verified (means tested). Finally this aspect looks at how well the components correspond to coding, design and architectural rules. Functionality aspect The functionality aspect describes what functionality that can be expected to work in a component of this rank, and how this has been tested and verified. On figure 2 legacy and key legacy requirements are presented. It is important to emphasize that during development stage in addition all legacy functionality must be preserved. In order to prove this, key legacy functionality must be tested. Component Test is executed for the component defined as A software component must have a specification in order to be tested. Given any initial state of the component, in a defined environment, for any fully-defined sequence of inputs and any observed outcome, it shall be possible to establish whether or not the component conforms to the specification [4].  SHAPE \* MERGEFORMAT  Legacy requirements on component test Component Test is executed by the designer or design team for the software component under test. If there is a need for an integration before it is possible to perform a Component Test, the software can be integrated and delivered back to the designer for Component Test. Component Test is executed prior to delivery to next phase. Exhaustive state-machine based testing for high functional and non-functional coverage and error conditions (during early verification phases). Load and negative testing performed to validate system stability. Specification-based and code-based test techniques are often contrasted as functional vs. structural testing. These two approaches to test selection are not to be seen as alternative but rather as complementary; in fact, they use different sources of information and have proved to highlight different kinds of problems. They could be used in combination, depending on budgetary considerations. Structural testing is used in computer programming, software engineering and software testing to check that the outputs of a program, given certain inputs, conform to the internal design and implementation of the program. The term structural indicates that the tester closely examines the internal implementation of the program being tested. Structural testing is also known as white box testing. Functional testing is used in computer programming, software engineering and software testing to check that the outputs of a program, given certain inputs, conform to the functional specification of the program. The internal implementation of the program being executed is not examined by the tester. This is also known as black box testing. On figure 3 are described differences between high priority key legacy test cases, key legacy test cases and legacy test cases. It is important that selection of the test cases is divided on appropriate way to cover most important call scenarios.  SHAPE \* MERGEFORMAT  Component test Code coverage describes the degree to which the source code of a program has been tested. There are a number of different ways of measuring code coverage. Code coverage is expressed as a percentage, as in "We have tested 67% of the code". The meaning of this depends on what form(s) of code coverage have been used, as 67% path coverage is more comprehensive than 67% statement coverage. The functionality evolves from just handling the correct input to handling all possible cases of input. Capability aspect The capability aspect describes how we have handled different non-functional requirements. Different applications will have different requirements on the capability of the component. Some applications will have requirements on code size; other on efficient memory handling. Capability of the software components can be checked using different types of the reviews. Actions shall be taken to remove known bottlenecks in code. This means the component shall be optimized for memory, speed or other area, when appropriate. This is based on measurements provided by profiling at system level. Run-time checking of potential error conditions to prevent crashes. Real-time performance monitoring used to trigger Central Processor Unit (CPU) load sharing or re-allocation of thread priorities or termination of threads. For memory profiling more than one non-interchangeable tool should be used. Spare variables in block exist for all types of variables (e.g. R declared, reload declared, per record). Unused variables removed from block. Event reporting/ alarms for scenarios not required through requirement specification is implemented (e.g. exceeding customer expectations on functionality), when applicable. an example of the definition of the sqr for the software component In this chapter an example of the SQR for the software component is presented. It describes SQR for the software component which is in development for 3 years and in live operation for 2 months within TSS (Trunk Signaling Subsystem) part of the AXE.  SQR table Depending on the results for every single aspects, and aspect contribution, the ranking is made. The overall SQR rank is the lowest rank for every single aspect. SQR level for the software component on figure 4 is 4 because the lowest rank for the single SQR aspect is the 4. Maintainability aspect is 4 because code comments contribution is the lowest in the maintainability aspect. Functionality aspect is 4 because regression test contribution is the lowest in functionality aspects. The evaluation and reviews was made step-by-step by experts from every single field (design, system management, test) and for every single aspect (maintainability, structure, functionality and capability). It comprises experts from several areas from the organization with in some cases more than 10 years of software engineering experience. On the other hand, it must be noted that there are always some limitations in judgment during the evaluation and review process when the people are involved. Two experts from the same field of expertise will always have some differences with the opinion, but in this case it is acceptable because on the scale from 1 to 8 with the raw granularity we are talking about +/- 1 point. Nevertheless, further work needs to take into the consideration granulation effect on structure assessment. The lowest level in Maintainability aspect is Code comments. Code comments are on level 4 (comments to support understanding/expectations on component interfaces). As a result for this SQR level is preventive action on the beginning of the project to increase SQR level on level 5 (which is the minimum target SQR level for every single software component). The preventive actions should be that i.e. all newly added code should be properly commented and the code shall contain enough comments to help an experienced designer with no pre-knowledge of the code to understand and be able to do corrections and smaller alterations in the code. If all these criteria are met, the software component will be on SQR level 5 for Maintainability aspect. The lowest level for Functionality aspect is Regression test. Regression test is on level 4 (optional Flow test cases shall be executed. Influence on other parts of the system shall be tested. At least 30% of high priority test cases on impacted area shall be executed on the last component version. The preventive action should be that Basic Test Instructions for software component should be written to cover all configuration sequences, all main traffic scenarios and all optional call flows should be covered with overall code coverage above 60% (we should try to reach at least 60% of coverage, more the better). Also, 40% of high priority test cases must be executed. If all these criteria are met, the software component will be on SQR level 5 for Functionality aspect. Figure 5 shows final SQR score chart as a result of the SQR evaluation. SQR score chart is made in Microsoft Excel [14]. It should be noted that the SQR level for last two aspects (Structure and Capability aspects) is above wanted level (for this software components wanted SQR level is at least 5 on scale from 1 to 8). As a consequence of that no mandatory preventive actions will be taken for those aspects, but it will be leave-up to project manager decision about overall quality requirements and based on the budget within the project.  SQR score chart All these preventive actions from different aspects should be driven and monitored inside the current running developing project by the relevant project managers. In the software development process at Ericsson Nikola Tesla, while quality is seen from all of the above aspects, the most important view is view of the project managers. The situation in which the software is developed and used heavily influences the project manager's view. To accomplish the task, project managers must have indicators for improving the project/product performances and achieving the required level of quality for the product. Thus the project manager's view of software quality is pragmatic and relatively simple - quality software is software which serves its intended function and is software that is reliable to perform that function [7]. Any model and associated metrics program that is to be funded by a project manager must be aimed at satisfaction of the quality criteria and at the identification of risks that they will not be met [15]. SQR as a method for early fault detection is related to prediction of software quality [13], where presented and evaluated models can help in prediction in number of the faults in the system during his lifetime and consequently helps to improve the quality of the product taking the appropriate preventive actions. Investing the time and money on removing the faults in the software components on time will reduce further development cost, maintenance cost and will increase customer confidence in the application and the company in general. Conclusion Quality is something very hard to define, but it is a measure about how confident is user of the services in operator/vendor. It is always about quality and how the product behaves during live operations. However, fault-free product most likely will not be affordable. Without some balance to the interests of the Quality assurance function, it can become too large. These are the influences of the classic market-share dilemma. There is no perfect quality; only good enough. Its time that we developed approaches and methodologies that apply to the whole craft, not just to space missions, medical devices, or academic experiments. It must be stressed again, quality is very important especially today when we have huge competitions on the market. We are witnesses that some of the big vendors with the long history are not very successful today on the market. On the other hand, some new players are very eager to grab market cake. Who will win in this competition? Manufacturers with good enough and first time right quality products. And how can we reach the good enough quality products? SQR method presented in this paper will help in early detection of the quality of the products and will propose to do some preventive actions instead corrective actions later on. SQR as such does not have alternative and it is rather further contribution in overall quality activities and software component quality baseline. SQR is a new quality assurance method in software development process based on software quality ranks by means of which should be achieved both, better faults removal and some level of quality prediction. On such way, project manager can take appropriate actions very early on the begging of the projects to avoid problems in the later phases to gain quality as we want from our products. References Miroslaw Staron, Wilhelm Meding, Predicting Short-Term Defect Inflow in Large Software Projects An Initial Evaluation, 11th International Conference on Evaluation and Assessment in Software Engineering, EASE 2007. James Bach,  Good Enough Quality: Beyond the Buzzword , Software Realities, 1997. Lovre Hribar, Sanja Bogovac, Zdenko Marin i,  Implementation of Fault Slip Through in Design Phase of the Project, 31. International Conference, MIPRO 2008. British Standard for Component Testing, BS Std 7925-2 RUP, IBM Rational,  HYPERLINK "http://www-306.ibm.com/software/awdtools/rup/" http://www-306.ibm.com/software/awdtools/rup/ IEEE Std 1028-1997 IEEE Standard for Software Reviews Description Hyatt, L.E., Rosenberg, L.H., NASA Goddard Space Flight Centre, United States of America, A Software Quality Model and Metrics for Identifying Project Risks and Assessing Software Quality, ESA 1996. Damm, L-O, Lundberg, L., Wohlin C., Faults-slip-through A Concept for Measuring the Efficiency of the T Test Process, Wiley Software Process: Improvement and Practice, Special Issue, 2006. Fenton NE and Ohlsson N, "Quantitative Analysis of Faults and Failures in a Complex Software System",IEEE Transactions on Software Engineering, 26(8), 797-814,2000. IEEE standard for a software quality metrics methodology, IEEE Std 1061-1998, Dec 1998. ISO/IEC 9126-1:2001 --- Software engineering -- Product quality -- Part 1: Quality model, 2001. Walt Lipke, "Right Sizing Quality Assurance", CrossTalk, The Journal of Defence Software Engeneering, July 2004. Lovre Hribar, Usage of Weibull and other models for software faults prediction in AXE, 2008. International Conference on Software, Telecommunications & Computer Networks, SoftCOM 2008  HYPERLINK "http://office.microsoft.com/enus/excel/FX100487621033.aspx" http://office.microsoft.com/enus/excel/FX100487621033.aspx \h     ( Z j l n    4 5 6 Q R S T ڷqg` hchfbhchJN0J$>*jhchJNU hchJNjhchJNU hchdhchI0J$>*jhchIU hchIjhchIUh,Kh4PJh,KhdPJ hcPJ h,Kh- h,Kh- PJ h,Kh hch>& h,Kh=g# l 5gdJ7`gdJ7gd]gdxgd/gdBgddxgdcgd 4KT U l   - ; I \ 4.8E]N=ǽǶǡyrǎkd] h,Kh)3 h,Kh\ h,KhBdu h,Khe h,Kh] h,Kh-`[ h,Khx h,Kh/h,Kh- mH nH uh,Kh- PJmH nH u h,Kh^? h,KhIh,Kh- 6PJ h,Kh- hchIhchfb0J$>*jhchfbU hchfbjhchfbU$=B`*;57BCݭ򊑊|ungng`g h,KhY h,Kh8 h,Kh[8b h,Kh- h,Khq h,Kh H h,KhnM h,Kh$1 h,KhC!h,Kh,KhJ76]h,KhJ7PJjh,KhJ7U jh,KhJ7UhnHtHjh,KhJ7U h,KhJ7 h,Khx h,Khf5 h,Kh/ h,KhpD$;7EVWc$%< !##$%L'(`gd.8gd:`gd H`gd#`gd *"gdjgd*YgdBgd[8bgd$1CDEUVWbc#%4;<HO !#(:((i)W*****+++Ľzslse^We h,Kh- h,Kh)D h,KhE h,Kh.8 h,Kh: h,Kh# h,Kh *"h,Kh- \mH nH uh,Kh H\mH nH uh,Khj\mH nH uh,K h,Khj h,Kh Hh,Kh- PJmH nH uh,Kh HPJmH nH u h,Kh,KhB h,KhB h,Kh8 h,KhAD!(i)**+L-:../x0Y1111l3444455gd'X`gd'XgdMgd'Xgd^gd Hgd^?gd gd >W`gd H`gd)D`gdE+///v0Y1111111e2p2z2k34444444444555j6=7U:V:;<ǹդ}tjf_XX h,Kh h,Kh+[h,Kh,Kh'X6]h,Kh'XPJjah,Kh'XU jh,Kh'XUhnHtHjh,Kh'XU h,Kh'X h,Kh^ h,KhMh,Kh- \mH nH uh,Kh H\mH nH u h,Khx h,Kh- h,Kh h,Kh HH* h,Kh H h,Kh >W"5j6=78W:;<<<<<<X>>>>B@BBCC $@&`a$gd.8 $@&`a$gd69 $@&`a$gd+[gd Hgd'X`gd'Xgdgd+[gdM<<<<<<<<<<<<<<X>>>>>>-?/??@@@@fAoAAAAABBkCunungng`g`g`gYg h,Kh69 h,Kh^? h,Kh+[ h,Kh(h,Kh H@hh,Kh- \mH nH uh,Kh H\mH nH u h,Kh- h,Kh H h,KhMh,Kh,Kh'X6]h,Kh'XPJjh,Kh'XU jh,Kh'XUhnHtHjh,Kh'XU h,Kh'X h,Kh,KhM"kCxCCCCCCCCCCCCDDvD{DDDDDDDDDDDDD{E|Eݵ뮧vmc_XQ h,Kh6 h,Kh*Yhh,Kh=6]h,Kh=PJh,KhbUm@h h,Kh_jh,Kh_U h,KhbUm h,Kh- h,Khe h,Kh) h,KhV^h,Kh- mH nH uh,KhV^hmH nH uh,KhbUmhmH nH uh,KhhmH nH u h,Kh= h,Kh+[ h,Kh.8CCDDDDD}EFJLOnPRRRR'RwScU/Vgdl94gd=gd{r`rgdgdbUmgdIagd gdoB gdgd=`gd_gd!gdegd|E}EEEEEEEEEEEEFF h,Khsh h,Kh-`[ h,KhYh,Kh- mH nH uh,Kh- hmH nH uh h,Khl94 h,KhA Q h,Kh h,Khk h,Kh=h,Kh=6]h,Kh{PJh,Kh=PJ$/VMXNXYX7Z\\p_{_`NaFb}bbAc ddsee+fgdxgd)Dgd^?gdVIgd >W$a$gd!gdx.gdc gd/gd gd-`[gd$gdYgdbUmgdp_{__Na9bEbFb}bbbbbbbbbAccccc d dֿ~qbQbQbFh,KhVInHtH h,KhVINHPJaJnHtHh,KhVIPJaJnHtHh,Kh >WaJnHtHh,Kh!0J$aJnHtH'j,h,Kh!UaJnHtH!jh,Kh!UaJnHtHh,Kh!aJnHtHh,Khx.aJnHtHh,Khc \aJh,Khc aJh,Kh/aJh,Kh/aJh,Kh- PJmH nH u dddddFeqeseeeff+f7f8fVfXffUgVgWgcgggǾoeXMF=Fh,KhIaCJ h,KhIajh,KhIaUh,KhNCaJnHtHh,KhNC\aJ3h,KhAD0J/5B*CJOJQJ\^JaJphh,KhAD5\aJh,KhADaJ h,KhxNHPJaJnHtHh,KhxPJaJnHtHh,Kh- aJh,Kh)DaJ h,Kh)Dh,KhpDaJ h,KhpDNHPJaJnHtHh,KhpDPJaJnHtH+ffVgg45?@ABCDEFGHIJKL$ & F$dNa$ 7$8$H$gdJ7$a$gd#gdwgdIagdNCgdADgggggg~Ӑ345?KLȿ)h0nhJ7B* OJQJ^JmH ph2XsH h- h,Kh#CJh,KhKYg\aJh,Kh#\aJh,KhwCJh,Khw\aJUh,KhIaCJh,KhIa0J$jh,KhIaUj .h,KhIaULovre Hribar, First Time Right in AXE using One track and Stream Line Development, 32. International Conference, MIPRO 2009. A. Richard Newton, Component-Based Software State-of-the-Art and Lessons Learned, Department of Electrical Engineering and Computer Sciences, University of California at Berkeley SQR focus 4&P 1hP. A!n"n#$% 7 0&P 1hP. A!n"n#$% A 0&P 1hP:p). A!n"n#$% P TFUa%Sa6`sJFIF``C    $.' ",#(7),01444'9=82<.342C  2!!22222222222222222222222222222222222222222222222222^P" }!1AQa"q2#BR$3br %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz w!1AQaq"2B #3Rbr $4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ?( ( ( 5>/Osw}s.҉ 2lx^v3gq<o =d~O[j4=jZ-J%@{P}9zЯ>x&Zx4r,}pq0tQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEp?&4[ >@(g3u{Kޙ&EkkWsEXk/1ƏqI]<ynuPqIGCGoo0xkT:9xGov_`5N"2:FeaG%x:Ꮝm  f_a }3x>>Sj\/=A> ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( -O/dyuytezk"& G&kX%$u@~aFԴ%D팕S c#K{|4luY-nIe\295[öՎf3{Y#Y$ؠ((((((((޹.‘Rvw*c }jKФyo-s.WA䐹 {OEe4XtT{hǞx((((((HyIX>< T4GსDO {W_ZʹԆkMOwkEBOjvz\YMFlw=q^|:妇]wvMvk' Ԓ;D$oc<;YxCϨjai1Z|I Ibsր=Ku4YeIx7epB+FQmgTaԮVӴ1VZТ &+\ZVqtei*!ր;{IHmBK#aQ@$TQv?)7>B*qג_6"F x4h7\1\HS^K{jgKF'$)Ga@Ey—mjwUL.fzJ'aydTRjMIu]FmRp\\V6(*L2s@Es"_' -xg ,]CzG'/P*+ZH<|C;Zh}d{(kIr3>RA QEQEQEQEQEQQ\Cim-ċ1!GcI?A@o=m%۩ꁡ7gqׂvE7btA}~7p:O+i@VT&`V[k/p"P0}HzvMҬ5(--o.Y`w0 rGŽtۨ5(fp+8I蠃N)ڟ =GW@,(@Vf-IR as >hq^^]1;H p8#S<[:hzwY7~e͎<"/lUI˻v9sv݋mV+{jR2bXs5PWzFK%舃%ISu+m'wqB6eزYeP&V08RFקQ@N4{NJYWkvד%պE%qƑ TSԩo/>{P%ksMIhol6#zbziqd^\C>!b.r(s "8rF1HO@ǯ=ʆPt1R0A;=|g}Pu 6O3A }+ҫڿ^iqW4]7#8?"l<[]gOl:O=f(((((] :V )h z~i+2i64la>@8QESҬ5 KHO6rcoQ{ W4z(b1%^2F Vǡ*(:@m` >㱐l3dt6,2=lRE,Ƿa?MU(>}K[Vi)>L P,,οu#vǵeoqL96Ѣ𵔸eSB #S^y_ɲ##'R+ W~*K{6'گ0N pl_"(1o}6׮j"ƊQ@ 0<J5K/vWg7hJBܤ*5gK`tk^IL /J*@ 85PZJg۟XXMQn򵳙Ex8#}͵>$u46KD[c9| x&NHu \jnw*ŽWTJ6m#qK;oV_j@n3ֽ I,5-uc}zn:ޮϫ^GeS{[YNﶩUdfB2z uZdv$siz|*@Wpvk$u46KD[c9| x&Naw}o&TTIfv"?WǾ ^`ʨJaǠ3Z(5{]CMλ{5ֵoOKbh1VH wuâϢYL_ҮF >OXO?C9tW???AEsO?C9O?C9CwRqY)C=5|.΋+a,MMϒ(U9|a?k/65#'|1u_hl,Y @iA7xF\^?y$z^?2!sNdW??j߅u+M$ 9FΒ!Hg<s(h7h7s<s(h7h7s<s(h7h7s<s(h7h7s<s(h7h7s|QxJ[Ǹo i($wzQ@Q@Q@Q@Q@Q@Q@/|a $^&ЃL$SezFG#le;Ejg)'?oϏ$ n3s0}~ѿ4~tj1 =o| .4 Nl1-qv^>QEQEQETW26K\:+eC9^K$d=+c_5-ŧpbgգE^=-%R23P:eZ_ڎ o?OdYy#>#]7Ou]6qw ZʇyS{f"Z|QAJm&a9۰yz7V HHbR:l *3(>Aޟy"k!YcиDhe+yep1Yu/jtb351 GO7͞h&dae9z\Uߍ-ZMi2}E>wpZ‚8 (acO0;z @ d@'ԣ]:ix02X1%„tg& oγuCh1}6,eF4Fw݌FqY=/_jv4v3? D3 *3e$ JͭAhwPI:Gxdϱdd}׮9_X'|/c]hc R~#W?,o}Z !. L2pEtuOWK Q7.HtSzGmn`U IrH# h()4RƲF 2(iz$sG[G4lR8Q@/];SIͶv~Cj,GbȒ4rFHC ZTPUk8,a4F X͒sԒh:d%Kӝ[|׌m^֍q&j$FO Lx|MfUiR[- N:ޢ0 +O,;ÒIc3R[x[Dk_ hw3l;7_Zآ0<%bO`64I#yGaUMi[pŨu矔z(zVS8ʝ)aАx5r xm`HmaR5 =( ( ( ( ӬgbFI#]C7Ţ[̎y2?0?^BxPJ\KpץOݏ?"~$d]31ERCc9~Ҁ=2 8--bXE8pMEQE~v1ҳ#@ 6% ‚K\IE!U,,:r)hUW;T Q@Q@ʭFGzUk(aFih((((((((nmx^I\*ԞfY4 kYdB 2 C ?0i_ g`ҿ6?YϤQ ?0i_ g`ҿ6?YϤQ ?0i_ g`ҿ6?YϤQ ?0i_ g`ҿ6?YϤQ ?0i_ g`ҿ6?YϤQ ?7cii֒u&0x{Gz~LLυu۴&pђx} ?ãhyvS>ԑ yMxo +c?, W5#񥇍;^Al5% rZ* w~ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6? lZ`H?أ g`ҿ6?:Ϗ5zk:|ggU'g>߱_8|{T7EfLH3$pvs_T=6Swv3*+~Xqt50_~G5|2<mkum7fP9s>0>=A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~gY4 h0i_և,(A~P+ J$?z0r-`bPEPEPEPEPEPEazo QJ(C%@#8(unú-ƫxw-0QHGzҢ(nsٙG"a%ܡ<@W4Vk[rÆR0EX>FִcĘlXm#}ۘ C;WԞ t McNug#wV_xbm6]ݧ-'#\qOC W|.o\h:x4nwp$ݎ:p()(e w(((((((((((_W5f*n1Z+}#fW'>kk |s\GoU{|@A3=dD3՚Nӭ_J?kxU 1M܅(zݵ|!k%M%1@I7+.휟z TI. z}dG>^}(5uFN}v-BHb Xg sֹUMMm,$>!x[2@泵hHTt?|]3OXD+cl"#"]A\cyǭy֕jmj2ܑ5k]?O5} }km[w91FMz9Оowy祼M@\(vՉzkBm>+帇\Б@a:o~-x*z4YݳGGjo9߆#]#[Gj((((((((((+~0Ȗ5 k |Aܭz$1<:h 31P:kj7>+Gk3{LRKHG,@6y_},C}>} T}*B4[XQ|Ԟ&EPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPEPnC5[*ު{0 |auNH'dzG}u\og諫ق7Oq@Zkea:is)nng%O -=i++e6Nz<C5坾g-BX%]P> WU֭zy ?Tm4ftKa3B< W5 <7Fҥ<'΋c|+Z[ ˡpC,Jc0zzEr:7euk YFڭ27 =5}|`O#UFS<všm؅uXJ }m㜒Uq]7:iviwx{_\l]@ƥ m5Fd(J$G )p}C~mKkiYm7 B0WC+n{Ku\(Lqf(k>&eOm\K6R%R#-m! s;ny;ᷖ8FػHR$Z?TG|m^Qkg\R$*8<@4}Mo[cc塐.!lۀN9?ºuն<1&0m|* 91Otkv=aK#4a*nA$ c3YmS}v[i-[32#DrUب$}9v}=-a&⍊<cٰ3Se ~}aqgf l€(U8Ȭ jimy1mSi&fbvTq)uk#_십:;6w} (y$P/;+94E"ilfTpS0WI\׆k7iI h .xwO*/-V;m*JdG!1pH֭g@[daJ>!|'6ic,g[ŖYFXGNJ}k?߳߆4Z냂LYuNz=v;hG jv+]KԯeZ粁>_(((((((((((((((((((((X'Ԥ)lVRU}x C4W42$HсVR2#4([-̅ !mA p8O'EPEPEP~/&~S]goڔ, yp݇QmxHeR2HwW7֧ڦm!c(Hƞ,NlГ )޷(((((((|mRv{H5,%1{]ॺe# xUS_6,_zf헌É$'(L_ KsuOS <cOv>Q@W ֗s/on`-4rH^T3MvWMӬUx ,?x sZW7lg3k{bi @ր:Z+P~cvCyHa[>E}b;qU'͵y&m{V7 - lMRn Al8$Y:n_[Oxn3-X>^hCf{uo{ ^Z#wGHPHioi^5ȷKi1>p}yh[^ŭIymK7[ϳvGic<i&m6s+񕎗4֓G|*Լvhb_8;Pp INӦx㷾@cf NJ訮^Ow㡣Cm#Xw#bNI>EuQEQEQEQEQEQEQEQEQEQEQEQEQE—;"ºdg (_ɒs1\6}c[Fm:/Z,K7$pR21g%{4+9%y ]wrrIS'hZmYE4PTq+@$~nWhLN$V99!OQvQjM C4,YI}zWh|-t;Cr~l:z=|Ic%<#'4{q7xQRY`A!ZEp7prNMYQҼo wS}.$W ĒW8`I#@/nud-չQ3*Tg9O5SɿgkPO(m⑱@⤤U T`ږ ( |g᫝n`2a};g|爧^%%͸<YOj^ ( tO wNŗߜ^+8MwjwvpVZ q+O rҲ<%C}2hW:%7P$(]dY*uPq^Eyմw> z\7A< yFEc$@|=үqUŠОi׎* vʒ:]䑌#[G,*򪢅P x[ΕY^]rCrڦ_(IU!݀@e/Aamkjdo=Z?)Y' XP .h-?O҆<0\aiT+,l)CdzU nC&׵M&|8e?uPtۭ#n{ȡ`'ɼ台c3y۱;GEy mO^t[Qu)e䷊6u Rhċ3 H8ڿoa1g35)d26APizퟌl/bKF$ dffm$cA]PEPEPEPEPEPEPEPEPEP7fm$6I Y@$~Ux|t? WAEsw?Cɿw?CɿtP?x|t? Qx|t? WAEsw?Cɿw?CɿtP?x|t? Qx|t? WAEsw?Cɿw?CɿtP?x|t? Qx|t? WAEsw?Cɿw?CɿtP?x|t? Qx|t? WAEsw?Cɿw?CɿtP?x|t? W_x{S>-p żCrT1c-UԴMcLӯYnc1w}{P6Fa CمKHM`y0&U5SZKn, wB煺'>Wc]We-nSrЃ~G77ώ<ώ(77ώ<ώ(77ώǾ"'xHXC/#vh|dpquE{_+n-7gr?ׂ|f}_< :/Gq޴=jŚT9Os,C:nr(gN_ jgPOI4C}G> ( ( ( Kinn$X ;P2I*JOٺ<^7̺*yHAD~J}hoic[M'-S\elW}[XY"bჯ^EUCGpuQMQEQEGA bs9*q8 ($($K@Q@S%(#i%#DRQԓ(Sb9Ibu7Pr@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Q@Vm7/KhUv/%wM+*@xf*ƱXu}ZMu-F( {9U2rE'+F}[T7֌@̦uWݥNOA@*-u)xlsNm! ~CXt犩xWbE=FTȞxlw>Z* ^h2{!Kw_z ].[qXĻE}>O"R+g~nBxKG0r'g;s]OHd{+a.3L1u.l[@ihO`}~6ۻۻn{Pu5KTm F+[c;LSѹ?U5۫^]0ޠn-Kxϙ*d``mq>!IXZ!<܀h\ȗڍn$WF Rvz]ŷ/ 4ihɀtT]ׯI=yҚGrF8_Yyzq,.,-;YT;\2n1H ]V @s.>v_z#]?RK!*Xw ѓ2vvjUU[PN$,lHl {pkOǘMs75#~Yl5+;}-V[A'!2UbH_:em$> LWm4G퍱{RjZʼn< ա $;z<t67w:F mH`LB~] 0bZgS,Jn cMsčoaK+h%= p瞣k!ln-/UJ0ǹ ܁].E5>XmR崕U>TG8`<՛ ize֛=rEk% #dųbP#^]j2$Vwټ[o1o3JRm̄\t8w98jQEQEQEQEQEQEQEQEQEQEQEQETXZIur!ei8Aco?(C !dhM;A'G&A7? MwO&: +ÿ24o?(C !dhM;A'G&A7? MwO&: +ÿ24o?(C !dhM;A'G&A7? MwO&:$@L[ )S=6qs+M;A'Yޯ5`=[i SnT |C'V [Ȧ'\c_ X|˿ k~zu n{#6>Eƌoo+mC!(F~^4Q\&AC A\So 3gv?%<*|`M/&A>5Ѽ]4]5K "R< (((((^궚}͍Õc >U'Ͼ]2E;ZAQ$OEgž&o/,efGS= (AkhpۏEe4mP'S꺋ivF,//`M+?=Hh^:M^`]٦H# N$$`0'ֺ(ڥ-h%H0km k1 jiY^^IzU\h֗>gbK_f%S1 ^-m4[#Fުa[95^CȮ.MrЁ"di v:@s .GBx;ƞV/(ǁ#(B><$>l6zdV]&YZ8%3XU,A>2ʭ efX ?׀Ea;Rr`H'; D?Yh..d[%($e mb2y=k{)izuƞt0;2̲mN1qZ[̒:JAYF0OA \iz.`J doύ-UW$!bsV:{7q69X2I6J'`Nȹ66'%08j{XY= #C;Tp1(5-z{_skfr^I=uvnҩ^(ǯUE-d<$[F39nѮ]u4V$8|L-'C^s^,Kuem:s%`#ʏyihݷ3ydVQXw֋%> 7lAZz2xM$p QDQTUb8k:fx^I5EmfY<)73m',Rk]&{|V]tdm}j'+׫z~d666֦S o>@b1#1P9q~zWԚB;cdq x<"?#gy{1n8;b%:JAYF9#A UQB@pr>2y+L58Ƽë[cɇ<<̔ {V^lo-oΉnl'WH-c޽f[xfhXcmYA*}G %$ ::@dt$|z6УYVkvLv y~wq^ hʗ6OAw,enAr@]lC,- #yASUE QEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQEQE@=ۨv[ߚ`J:c 3W&\sx՜ tEHn$5y<51  B+L`! ,XD[ŃJۀiOHThQUlJDnݻ} \w@Y a%b#+!;yIi&x2h !o}֓![fT@C!@u@>.΁p+͛4DZ_3w_u{STR>!K39.R.B\{1V?6h: >֏$#pA4sG&׎InGS;V(ͷKI'ʼnco*qMׅc~cww4by,S3 j;vĺwYVH/{F )|Wc{:pB͇eLLnL yh:Hd=8$(ch:ʰpi Vոg@o!q*&6 >^YLƴ*i H|iYw 4VBGBk!xi8\Hͳr.-}= GjgUwAԎ0(5Gc_x?Pv="{)s W.xv Yge1Th!XP ⤝#kS6֞==Cg)Om6Y|x/}' m/{>[Td{"l?d sD3n+x>- LorJ^n})JL`ʰgVl| 屄V.bjwp̵)Cߌ kMoz?1MˡDt!}2]ѿ@f7`6[pi=/S]p-O6~sqeN7rj"M:ԪSk?szr#^ԛ ڧ`z*`oBdzKUB`հd喗 QpAQp& Z4ZBphg~} uP-nheok u.c(ƼId1X1C|}}ى!EMm;(Mdph չHWOH/l4gE>k6=8=pygTC^ίq:ﾰ"֪#({bRZ^emLTn"Y:2]o]}aP!0zvóA:`' b1㗷tPLH 2*ԅHP ܧ^|ۜ%Y:9e\B,9;eUry ܓ4lg鰑ޞnHJ=ו .IՕIs~u>V:lג%6V$Q1k+ŬRs]N:F;;9i93748&tTjgX;`N]اS_?bVجAQ*yY"x*ZShE08' ljqs~pϣ7 A8ʩ,WvDA )evcm<%d'C_LTw6p}[0޲s+~/}Ni -n8B(28 p7/)$ e$0/) lt$`Ga, ˗`tagKpM?3:ю%q u(__A'{VO߅w (clԜE#k`sA1ls [s %cDN9a̟H d[>up?,'qr۠ؖџ!Pg|%3qsdY>VGfu|j<!Qdsצ,6a33הf/ Οu)}2bCk}I/fYx P!V= V=:< 5k_ߍ%Bp9EnrmJ69InBp>M}xBH(}2v"S`g*~4(5g-}GQ4rl,cq7k #522#epma-3kC4 ڐ~f emt2rk$Od4eϮW|a Z-6.RWbZL&9 |bi΁yVǣ'fl6צ\fs0a6'9l2?fw0ERd/3F7aAosJ=4jp_ڢ Q2|YX3cƲHF,Ab9Xb!Ѳ|5,ty(u:,X|,Pt(uX>*O(7 1˼zU[ókj}¬:2a0y64cݽc~Xh>.>糧ǺUۻ:u>Ld 1 c6 \L ^\i.XiER.{iȫV?ZkYz߃{[-ꁿDE}PکAH^) ԯuOD?){oRnUUZECK 5(諬H˪ߕÑOVѶ }C}rG6z˖w^\(pJ$\0+ rߛo)b.Вa@;Xn&$vLsx}b}q6oyg]IҜ>ߟO~=6؋s~=n lՃT-q?Q~jUP;p̭l۟;pⶣMmzjzw O:F~=?><&~=x8e1H2!<)DO/-N, DQ&FIEoJ5>f0C!A;ye6sichpp從ldE1`va y%f#:u8X@= - )N9 QDGR>  Ei?Yx\XTU?J%`wC%v 4`j(:E:?jP_eRٖR*dY-V~nfe_|3'޹λ9SYaiu/WMb/ zس#x^EWrk[ "Hc9!MD}&̌/PW9 ie$ʴUfL62NDζ s#+cza0Hr?FQ"a3{Xgw^v/?FO|AJCv?޽S!}rO9ۉFIPdC[L2?hgRm6 'ȧ"D-vNc` 2_tW_(YK$HLWi!+o4 ؚ$z A9&og$ǫ= M"LL̙=5}AAGb(3teueG [IG071߀P9~8eq_i]9/\$v,t Idg랥ڗ-K/hOw{X\NTmY \9Xcc-pU`A?k1-gmN`(?ҙ/zt2,|oRul?7 Q"uqrW0DFb;nkS-KȹNSi<=.h!WO_Nl&>֎>fK[-G3-fGEqz{`Xɼ-fl[bvl[r4sXɖ4S9 >Ec9줍^SςPdIWv%e ,:f#,ia])bL fUI0[7&m,@,+85cwlɿk(ͿwبN'-ƃ/(VaflIz_,Д pT|zD|OyLci[Fjw>4BE8l*؄oM1a2J Z/VprO"髩f Ap LY ]= t"m: h LQҰ~0&5i hVb=S}d9T,{*ώs.G#蓬u%nՋǭ | ɫW3OsA@+P ڡЬes nIhM gtpKm[nIWGc<6^`cEUɻ-0kEPی[Y0[Y`\G:,VXJ^U֨6uܜ>^VV4Υ>N?Ju}5[{ֶџc: Z>Ac]gwvEOlP[<7l'yHYQ́g:X< aD6D6w5%HVsam݉ǙBxt{=W`y & '4ts,BU kB ýPE}ΕqzxqU0_~й}aG.gryEVԫ0m*G|yW =[_[Tdkw3Ǝ9􃼙C7LSE5٧L:UTaJSyMcLzNq-x{D=OwoOP{sve>SGa-8Lt =LM 3?(3uHo%g~Mc}؟U3=ٟfnNusA{؟؟ln4 "Z-;Hy&o}֓Qώ}{v;CGsAٚ?wswLl@0V۔*=M9{sM\)̦4iSfS/MI>"8{5 Rhzd 8 S.r~z:X$Z]Nt6+/7IOo[X٧TD0j&?t׏GݶR[ Cdr~kQoۮp$_>w1%^ C.JUzѳo\ZTS}̺'SUFCOĿ~8h/ď̏UEOOZLJ[^!Uډ7w DSv vNW@cgd0` |GQPӄ`KC.+_/s'ɷl|M'ljJ:iqI1iΉMcnYa2Gq[ʦlaKe+yɷ|e9?297zڈol5^u|'6n |p꥿=~t_1咯+DNV 62Hoxtrk ,| SCcsU;iQQg[*LB8_q"1C+_ T  ՛|C|:D Zgo;9ƌ d:.}myeO-^wtv{k-OB='s:Wwg 6by(W~{c,@OOP#O5j&ZZmam'bCvXhC3JR~' N0S+Lu܇U h:/>h#ݏ632>ޮ'|Xlt@?VZ="{N^t:Sv[!ĥC!A:RI<P*7omkpol |!b>Eh;qN]5WQ; -=Wl Qp[G~IVG Ŕ@pa?3qA.ձO.L>pcpȳ\le*[,݌)2gG@f˅KnA2 .3 w , iژul{avvqo$z {SNh^CЊE .˜M/8KsPZ7Bm;+KX%ۨFG|yj&>& ьM|Dc$&kјw'S1c]6sgMl{]%Dd̾ +Ft{],CKcD|#2y|#|_AOGbݹ!]ΙA }LaH:!XDXxeuS?{4X`]3[X0%(дkf.=|E b0~' kD:&|U`ңLl6k96E{><ǏNjuB?I]'NL* :-KTuW-1/ v'V67q-n^Zff/ش̧})cY͙d^9 -1[hpKlz7#肸_Xlx5a-(jkxo5^Gr˺y,sx$ao1Ay tqf!1Sպu!y~B=# O{Ͻ'_80;1~nJ?1"hō4n,`0F?'o%L>BFiG3 |%!q.͕ٕs`)a-O-aO~- {?4֟~3Y{/DUx"٥ Ӣ2ũEo[Gʻʬg>yy;^զrJ_/ ;=aaݞ;}<9 -^5:餸laPɳ̒ (IMv)ׁe@{}1c]uց~angP쓊uw;xֆ0D4,K6cF#ILktI6&edzq3,cypjf3>'4<>=8gdL=OҌ{ ~bм#P/)mac*$wf:ٮJ۴|l-{<ޅ3 ^ mp{Gɣl3ۨwCp%QgJ6} z*1ęm#-H6v}s)(Sg@g`:mQsߜJ?޼~^Mi/Kmbot惾U+(8;Q1%}U6ӾW ڝ8G{̾fSfQו\o#<)!L4#S)8fl爧 1"rn_^s-.5j^~*GD&s_?:*&e;uޔ;2sGuKNf؜xr8b"y gCCiʱ?h<S)ŧ_ʧm\c!}W6>|Z*}/%G >K fSЛcA0j`#?"tl}v5`\Msd?޶߬܆eK3,疗2[WP¿۹wbޭL%E:*RIQFwfosKR?m`جm%v\}zHaK8ZMI65ی*P~Vz3鑮ƍ\/~=2b`jk,,Jj"z6k?ظw2+> ۥm)agҪT ɶ85BRT.Fg ؖgD_6g/1d%~ t(6iq6OPŨ1@Ŗ9 VإتH Bقơ=& 5ZOUf ԒTFZ/yQ %NzݫxygN[y طqYܧV|lBJ_ѣqt }{ \rMz(?}t*. Pŧ4YΟ# E>%x6{*yDPZк^Լ W Zr;PbP)G[~P-W.e3wP5æV$]~RvE&쒘$EHgQg¹iU}ރ%MzM$#/Ar ?3lVbɴvtf  \4k` =Q>yRĩT%YRJQQHso>mأQ(Gq<ԅri589z@մ,/^Vjַk~*oa( Ӫ+Duhج!qoZ(je*)Z#dm$.L©IS'ϒt0}d^;a^CUge!VLRsZbnN~ .~P)L]Ql$b91̓VC۫a'ݖYOTzY!I,sp-+>jK8xYWFy6.sMۤ<4)Pq}BзK+|s@uPYl?<|qzt._t:#Gs2f?3V')c[9Dd G9$#X0  # A"ZR41_bP{6Em @=.R41_bP{P@`*Wx] Tՙ>4C  882NPfK 002`$ hQ1J%nTdd|lWL*%Y?ۧO߾}oO7TsϹ=|qmBz:QKȴ]y&!s# aB"$zB A cI 5AqA ]<_1B@\ IDyy㒖֤e\ ɐ3N_HR:& i zBJv\E olV 9L!N6^Cnl>8:QM~sN`87xMA*8$PjrSv>HZ[?LͿ&3DZB.~n9?>-}LyhwOTFGLOHLxmrf  Ih ݙiwJ ݣy{uE˧-N7i4{b)8&6gd`s&+6Ay'&"_x< li &~) ׯs^ vJ5[U" v%J~P7a+L9˔tmz7Ge>8*ݣy{/| ti=Mv^EСOxWb( /Vt4/Yp,2ưl8>n/ Ȕe_(S'm~ nJ9ϭc_ythB? )2,'GNt%WBW7m7p4Ec$`2\t,[?ʘ.eJ+y_Mc^=t>e~}"yvFŽvT}>OżdtonM?ӥr<J_12ӏ.;\a uy.*%;ʱ3Iż,!{th|XI JT@V=*SȂ]d!SYc@.+|cEOi_b^KqR<auh@ ::$W֧|0ܫEĽ*#UQG"xeUY1J*yYtHY(yY~}#}FY{B^}đ6`+=u2$ yo@|fpij kAq8қAe 0YbbH ,˿d5(6y\LkAY̧ŵyӎk*DY̧ŦǦ-bb+;<6vl!6}0qa-;\Z\ؽ{>h\׮X}`:1!`>R">n.L뢎Ա|m;)e48#幉ww|yww|yww|yww, 1RNe qW˒g>BAKHyޗ͏oF$^砓_r ؙ\tgR𰮋Xܜl"kHFV-4VF glMdH3dmkҜvqp브7pi[6kزl˂,fK-Z̖'L[qYL[-[z\`ZŴafSa=[3[δpySL[!lZӖ>nmYT[|'g!f˗CұeWٖ-;5my*l`KeG9. f˛-lm|0Ӗ]2Ӗ`ifˉD^yNRĵun͵uno˶,ں5mykD^ٖE\[|\[i:嵂ں-ںam_ˢs[ks[9ز_ˢں-ںamw~A|~Y2 ߣL"yϮl2/W+e>d׽,Sr9U)sŮfe)xlS],Stb@.X,L2u,St2 v]k"LyH.Zvf)x:kV-E3|l5i\ǯ)뉂K{ю..i>~?#{y̫/bF C~^M7Zqմ}=?&^joB?#=_7D|I{!1?_4ZQs 5؏dQۚ`y;b#$mXc$B:<m[y#?~Gګ3 m8~v~.;^A qNI-ſKяu̥C?7:0ncj} 0]7xfo?lkI$YU_uw)Ew"۾\_'&-T}V cm-S|W=,5<z.r7P2ܷ9p1 >|'c'A+ 370ebe>X?c0?[Py^H_Om ^ ^݀]-!6RM.rpLϏ N|NۂI`9O 5R.׍Cm|poEX.b?X-yN\gr=Vr̮@w3cy.Zח@`lJ8}rm-|NoS|]S<}XEoVd}'_>A}|nO"/+O^^9bq7( 8BsJ.^ T |gBǯ4_~uU;} 3ӯt&uFEW/Ϛ$m?'>}ZW+yLgЯiW A ̕ y .~u?4l_ml2G@p&J8c]DrܤTl,υퟏst܇{?MLF^Ňqf׻?|6Wt28E:_SϩxMA "(ECxp:EMk[qJn8wxI6;ȕl:>}~'Av^AY+(m+.|i4\?#|@Z׿0BzT]X=JC}k/]P6ÿ*k=t~t/up'TqI:׬>JF+2}y]|Iw<]u7nvG|z%ϻ| WϯUlK4SIy,Т\s"4oJ(c?̯`zԓ{>0o!Vz;K-H׬z}Gパ*]LB^qi;vRe:*.mNYU֜|y7z%uyKqR߻[d ſꤽ:z\, Qe/U}o4`^%jq7_FK/)9;hy\d׻aQ*I~X|;HqH.EL89=p y=+|6` ^=T.1T,$LK8C5Rݮ:͕: '/ɿ zU\fC1;{L<> x+#Ia9:1y7<~yH6|gKoؒIo<qǿbZ Lpo {;.gOc˵G\9q꛼7 |'^kX_1ݛ6o(XF߰Lo<}&%M^0 ^< c/D _uGok+޽7߸d^ُ op p!RpLpljI؏ <ı 8Rvűz8v8pl±NkvZ4vl!#cH1Er'q ;B8q̊c Ǧ%wvX)SG{=p:;{1İjM8OJ Y bQqlX#cz8b8ǖ/j,_-^iq,'q "  8,!o+"MpL{qg؊*=COK .2;p{N8Ve8v0±cC mE´8,GC ;C8f8.!"21UeKDZIQc?ycJUTfش±Wc 6ĽW86D~cRyñ>0`"a>VtDZH8Vc!q$c6Eek.;K~خx T~c8v[;;4a]U8&xz86pl{脗JBql?yqc);1t biF'Ʊh8VcaH*ma,"z/X:=g8KxcVccWI0;Z!3HXK0MW! 4nƱX86PÆjq,VB8hc'jS8꽌c8viؿzc0z&O:cx\Qgj-(+Wk#Jr];Izw>$iOH˿҆휯N~=wT;WV>L[8 2- -DyK .http://www-306.ibm.com/software/awdtools/rup/yK \http://www-306.ibm.com/software/awdtools/rup/aDyK ;http://office.microsoft.com/enus/excel/FX100487621033.aspxyK vhttp://office.microsoft.com/enus/excel/FX100487621033.aspx0>@> Normal$a$_HmH sH tH j@j Heading 1/$$ & F h@P@&`:mHnHuj@j Heading 20$$$ & F x<@&^ `a$6mHnHud@d Heading 3*$ & F hd@&`a$6mHnHuf@f Heading 4,$ & F hh((@&`ha$6mHnHuP@P Heading 5P@& h:mHnHuDA@D Default Paragraph FontViV  Table Normal :V 44 la (k(No List NON Abstract $a$5CJ_HmH sH tH HOH Affiliation$a$_HmH sH tH JOJ Author$h(a$CJ_HmH sH tH NB@"N Body Text$d2`a$@hTO!2T bullet list# & F h^`FOBF equationd  lORl figure caption!$ & F hPa$CJ_HmH sH tH tObt footnote5 & F h (A&#$./` CJ_HmH sH tH ZOrZ key words$ x` a$56CJ_HmH sH tH VOV paper subtitle $xa$CJ_HmH sH tH POP paper title $xa$CJ0_HmH sH tH lOl references)$ & FhdL(^h`a$CJ_HmH sH tH VOV sponsors g$d` CJ_HmH sH tH @O@ table col head5CJFOF table col subhead6CJJOJ table copy$a$CJ_HmH sH tH ZOZ table footnote$<a$CJ _HmH sH tH nOn table head' $ & F h8dxa$:CJ_HmH sH tH 66  Footnote Text!@&!@ Footnote ReferenceH*RY2R  Document Map#-D M OJQJ^J6U@A6 Hyperlink >*B*ph$OQ$ goohl0<P@b< Body Text 2&$a$CJe@r HTML Preformatted='$ 2( Px 4 #\'*.25@9a$OJPJQJ^JmH sH h^@h Normal (Web)($dd[$\$a$ CJOJPJQJ^JaJmH sH NQ@N Body Text 3)$7$8$H$a$ OJQJaJrr TOC 1B*$ U"%S5$7$8$9DH$]S^`a$ CJmH sH tOt enumlev1=+$ 4P5$7$8$9DH$^`a$ CJmH sH hOh bUmList abc single line , & FCJOJQJ_HmH sH tH vOv bUmList abc single line (wide) - & FCJOJQJ_HmH sH tH V0@V x. List Bullet . & FCJOJQJ_HmH sH tH ZOZ ADctarticletitle1#5CJOJQJ\^JaJo(ph3f )_        d)_6oLL$:%%*_;00b0:;00e0;;00L0.;00M0.;00P00;00,0';00.0);0000i* 00 LL$:%%*_@0@0{L@0{L@0&9@0&9@0<@0w1@02@0]4 00bcd5 ;7EVWc$%<Li !!"L$:%%&x'Y((((l*++++,,j-=./W12333333X5555B799:::;;;;;}<=ACFnGIIII'IwJcL/MMONOYO7QSSpV{VUWWFX}XXAY ZZs[[+\\V]]\^______ _!_"_#_$_%_&_'_*_000000000 000000000000 000000000 00W0W 0WW0%0%0%0%0%0%0%0%0%0%0%0% 0WW0!0!0!0!0!0!0!0! 0WW0(0(0(0(0( 0(0(0(0(0(0(0(0(0(0(0( 0(0(0(0( 0WW0500000 00:0:0: 0:0:0:0:0:0:0:0:0:0:0:0: 0:0:0:0:0:0: 00NO0NO0NO0NOH0NO 0pV 0pV 0pV 0pV 0pV 0pV 0pV 0pV 0pV 0 pV 0 pV 0 pV 0 pV 0 pV 0pV 0pV 0000000000000X0 bcd5 ;7EWc<Li !!"L$:%%&x'Y((((l*++++,j-=./W123333X5555B799:::;;;;}<=ACFnGIII'IwJcLNOYO7QSSpV{VUWWFX}XXAY ZZs[[+\\V]]*_000000000  000000000000 0000000  0 0""000000000000 0""0!0!0!0!0!0!0!0! 0""0(0(0(0(0( 0(0(0(0(0(0(0(0(0( 0(0(0(( 0""(05( 0( 0( 0( 0( 0(  0060606 06060606060606060606 06060606  00G0G0G0GJ0G 04N 04N 04N 04N 04N 04N 04N 04N 04N 0 4N 0 4N 0 4N 0 4N 0 4N 00,T =C+<kC|ERp_ dgL478:<>?ABDEG(5C/V+fL59;=@CFK65QT+++333XXXV]]])_XXX___XX?R$%Sa6`s]a"$ۨv[ߚ"$ - )N9 Q@ &(  t  S @@A S"`?t  S @@A S"`? 9 # NI #  s"*?`  c $X99?9 # NIj  C .AV modell 29 #>Cz2  0R82X"`27,7I  Nr9Ir9IN2X >a6 N > fB B s *D2X8:'B=B S  ?+3)_9 t\ t t _Hlt205099716 _Hlt205099717 _Toc126485855 _Toc126485889 _Ref128796331 _Ref161678259 _Ref170900732 _Ref170901439 _Ref209587690/ ;FX}XXZ\*_@@;{XX@Yq[U]*_ @9t; @98 @9< @98 @9,@9"@94"@9: @9: @94: NYyYyYYZZZ*_ WYYYYZZZ*_ 8 *urn:schemas-microsoft-com:office:smarttagsCityB*urn:schemas-microsoft-com:office:smarttagscountry-region9 *urn:schemas-microsoft-com:office:smarttagsplace9*urn:schemas-microsoft-com:office:smarttagsState    r{_*__*_*UWc#%<,,333.4;;<IIjNNOYOOIRRSS____*__*_\c_-1Mn$RK.+.3 N?x!Y5&bi3Y)-C6f7+@O9Ԉr[b;l +<>`AjkQMr AP dJTRL-&eВX,@lX}-lpd 0!r~io+vlf,F?yT:H{H#©dM}z@pPh ^`OJQJo(^`OJPJQJ^Jo(-h pp^p`OJQJo(h @ @ ^@ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h ^`OJQJo(h ^`OJ QJ o(oh PP^P`OJQJo(^`CJOJQJo(^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.hh^h`o(-h` ` ^` `56CJo(hH.  ^ `>*o(- ^ `>*CJOJQJo({^{`56>*CJOJ PJ o(- s^`sOJ PJ o(   ^ `o( .....   ^ `o( ......   ^ `o(.......   ^ `o(........^`6o(.^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.h kk^k`OJQJo(h ;;^;`OJ QJ o(oh   ^ `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh {{^{`OJQJo(h KK^K`OJQJo(h ^`OJ QJ o(oh ^`OJQJo(0 ` 56789;<CJH*OJQJS*TXo( ^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.h ^`OJQJo(jj^j`OJPJQJ^Jo(-h : : ^: `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(  ^`OJQJo( ^`OJ QJ o(o pp^p`OJQJo( @ @ ^@ `OJQJo( ^`OJ QJ o(o ^`OJQJo( ^`OJQJo( ^`OJ QJ o(o PP^P`OJQJo(^`o(h ^`OJQJo(h jj^j`OJ QJ o(oh : : ^: `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(h ^`OJQJo(jj^j`OJPJQJ^Jo(-h : : ^: `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(-@`789;<B*CJH*OJQJS*TXo(.3 h^ `56789;<B*CJH*OJQJS*TXo(.3`56789;<B*CJH*OJQJS*TXo()h`h56CJOJQJo()@  ^@ o(x^o(()H^o(()^o(()^o(()] ] ^] `OJQJo(hH^`OJ QJ ^J o(hHohh^h`OJQJo(hH88^8`OJQJo(hH^`OJ QJ ^J o(hHo^`OJQJo(hH^`OJQJo(hHx x ^x `OJ QJ ^J o(hHoH#H#^H#`OJQJo(hHh ^`OJQJo(h ^`OJ QJ o(oh pp^p`OJQJo(h @ @ ^@ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h ^`OJQJo(h ^`OJ QJ o(oh PP^P`OJQJo(hh^h`56CJOJQJo([]h ^`OJQJo(h jj^j`OJQJo(g kg ^g `k5OJPJQJ^Jo(-h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(56B*CJOJQJo( Figure . ^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.856CJOJQJo( TABLE . h ^`OJQJo(h jj^j`OJ QJ o(oh : : ^: `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(h h ^h `CJOJQJo(^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.h ^`OJQJo(h jj^j`OJ QJ o(oh : : ^: `OJQJo(h   ^ `OJQJo(h ^`OJ QJ o(oh ^`OJQJo(h zz^z`OJQJo(h JJ^J`OJ QJ o(oh ^`OJQJo(^`OJPJQJ^Jo(- ^`OJ QJ o(o pp^p`OJQJo( @ @ ^@ `OJQJo( ^`OJ QJ o(o ^`OJQJo( ^`OJQJo( ^`OJ QJ o(o PP^P`OJQJo(h  ^ `OJQJo(hHh^`OJ QJ ^J o(hHohhh^h`OJQJo(hHh88^8`OJQJo(hHh^`OJ QJ ^J o(hHoh^`OJQJo(hHh^`OJQJo(hHhx x ^x `OJ QJ ^J o(hHohH#H#^H#`OJQJo(hH6f7X,@l&>`A>`A>`A>`AJTR-lO93 ?x!AP1MH{-&eY)-+<b;0!rF?ykQM\co+vdM}RK        0g        $                "v       0         z               A                 ꇸ                 W0x F)*+v3456=.!EUoF"wIq[;K""LW=idT5iP kolCI>wlNvw&xxnm{ - oB >& s6fbcJN{ x.wc q *"4;%\A'$1)3;M3l94.869^?NC)DpD HnMA Q >W+[-`[V^^Ia[8b28d=gKYgkbUmO8pSsqBduC!J7:\e" )VIB$,Kq!  !8X8vB?_d >AD $ cI#]Y=4*Y/E%'Xjuee4f5=(MB ct;ux4[d__*_-@ v 77 @WX])_PP@P`P@P @Unknown Gz Times New Roman5Symbol3& z ArialG5  hMS Mincho-3 fg;SimSun[SO5& z!TahomaI Arial Unicode MSkTimesNewRomanPSMTTimes New Roman;Wingdings?5 z Courier NewC .PMingLiUe0}fԚ=& hMS PGothic"Ah 2f2f&0P00P0r4d^^ 2qHP)?I2$Paper Title (use style: paper title)IEEEetklohrl                   Oh+'0  < H T `lt|(Paper Title (use style: paper title)IEEENormaletklohr3Microsoft Office Word@ @v$@rHm@D0P՜.+,D՜.+,T hp  IEEE0^' %Paper Title (use style: paper title) Title 8@ _PID_HLINKSAq6;http://office.microsoft.com/enus/excel/FX100487621033.aspx3 .http://www-306.ibm.com/software/awdtools/rup/3 -Q#mailto:darko.huljenic@ericsson.com3 y#mailto:ante.burilovic@ericsson.com3 J?!mailto:lovre.hribar@ericsson.com3   !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry FlData m/1TableWordDocumentSummaryInformation(DocumentSummaryInformation8CompObjq  FMicrosoft Office Word Document MSWordDocWord.Document.89q