The Use of Microcells as a Means of Optimizing UMTS Networks

Transcription

1 127 Proceedings of WFMN7, Chemnitz, Germany The Use of Microcells as a Means of Optimizing UMTS Networks David Pouhè, Driton Emini, Mathias Salbaum Technische Universität Berlin Teleca Systems GmbH T-Mobile Deutschland GmbH Einsteinufer 25 Neumeyerstr. 5 Dieselstr Berlin 9411 Nuermberg 9441 Nuermberg pouhe@emc.ee.tu-berlin.de, Driton.Emini@teleca.de, Matthias.Salbaum@t-mobile.de, Abstract This paper deals with the question whether the performance of UMTS networks can be improved by using micro cells operating on the same carrier. The crucial part is the setting of the microcell within a reference sector in a UMTS macro network in order to achieve maximum benefit. A further experiment will then reveal whether microcells are suitable for capacity extensions Index Terms UMTS, macrocell, microcell, cell load, network optimisation, handover II. COMPARISON OF THE PRESENT-STATE AND THE COMBINED - STATE The comparison is based on the results of several measurements in a particularly chosen area of investigation. The following three parameters, which have to be taken into consideration when planning and optimizing a network, constitute the so-called Optimization Triangle for Mobile Networks : Received Power PE (dbm), and Interference Ec/Io (db). I. INTRODUCTION An UMTS network requires precise planning and, when in use, continues supervision of all the parameters relevant to the quality of the service. The performance of the UMTS networks has so far been examined alongside with the use of sector-macro cells. Since the required capacity can be quickly consumed for quite a few services (multi media services, Mobile-TV, video call, etc.) network operators have to enhance network capacity in hot spot areas. A further alternative beside HSDPA and HSUPA is the introduction of microcells, which have already been tested and found to work well in GSM networks [1]. The success of such networks of the third generation is due to precise network planning in order to achieve quality, capacity and efficient management. [5] shows the different types and ways of application of a microcell. The utilisation of microcells has been investigated from different aspects in [2], [3] and [4]. This paper investigates the theoretical and practical use of microcell within a macrocell which are using the same frequency. The paper is organized as follows. Section II compares the present-state (with no micro cell in the system) and the combined-state (when a micro cell is additionally used within the system). Based on the results of this investigation the third section presents strategies to optimize UMTS networks taking the UMTS network of T-Mobile Germany in the city center of Nuremberg as an example. Section IV examines the network under high load. Fig.1. UMTS networks constituted of macro and micro cells (in circle). The first analysis in the area of investigation shows that, when a micro cell is implemented within the near field region of a macrocell, the throughput is lower than the case when no microcell is in the system (Fig. 2). Although the received power slightly improved, the interference situation worsened. Interferences are due to simultaneous data transmission from different connections within the cell and electromagnetic disturbance from co-channel cells. The increase of interference leads to a higher Bit Error Rate (BER), consequently increasing the repetition of the transferred data. Ultimately the throughput decreases. WFMN7_I_B2, pp

2 128 Proceedings of WFMN7, Chemnitz, Germany Macrocell and Microcell in System Only Macrocell in System Fig. 2. Throughput with and without microcell. Microcell within near field zone of reference macrocell. The potential of the environment surrounding the base station and the technical realisations in the area of investigation make it possible to carry out different experiments. The experiments vary in the distance and visual sight, Line of Sight (LOS) or Non Line of Sight (NLOS) respectively, between microcell and the reference macrocell. The main finding concerning the distance between the microcell 1 and the reference macrocell, within the near field region, is: The greater the distance between the two cells, the better the network performance (Fig.3) m (LOS) 8 m (LOS) 1 m (NLOS) Distance between Macrocell and Microcell Fig. 3. Throughput depending on the distance between macro- and microcell, stationary measurement. this means that the microcell can be placed at a maximum distance of 25m of the macro cell (depending on network structure and the surrounding environment). Since the distance between two or more macrocells in urban networks is 5 to 6m, the micro cell will then be at the edge of coverage of several macrocells. Besides having low coverage, these areas demonstrate instability regarding the best server in the system (which means there is no dominating sender in the activeset 2 ). Several equally strong senders will worsen the interference situation. It is also possible for gaps to occur as a result of cell breathing. With the use of microcells the performance of UMTS networks in such areas can be improved. Differenc CPICH (dbm) Fig. 4. Differential level when using a microcell in the system. In addition to the improvement of the received power 3 (Fig.4), the interference level is decreased since the microcell exists as a dominant sender within the system. The Signal-Noise- Ratio (SNR) increases and the data signal dominate over the noise signal. The average shows an improvement of the SNR by 3 to 4 db. Regarding the stability of the best server, an improvement is been achieved, too. In case no microcell is present in the system the best server oscillates permanently (Fig.6, Scrambling Code 4 ). This leads to instability of the server and high signalling effort. As a result of the gained finding, the following question arises: How does the system react if a microcell is positioned at the edge of coverage of a reference macrocell in urban networks? III. OPTIMAL PLACEMENT OF THE MICROCELL WITHIN URBAN UMTS NETWORKS The distance between the micro- and the macrocell has a crucial impact on the capacity and the coverage area [4] of the microcell. If the microcell is located at a long distance from the reference macrocell, the former is able to gain maximum capacity and the maximum coverage area. In urban networks 1 In both numerical and practical investigations, the used microcell has a circular pattern. Besides, it consists of three outdoor point sources and one indoor point source. Each point radiates ¼ of the total radiated power (2 W). 2 The Active Set shows the base station with which the mobile station is connected at the same time 3 The buildings close by can also achieve an improvement in terms of received power and interference situation when a micro cell is used 4 Altogether, 256 Scrambling Codes (SC) are used in the Downlink; they are regionally easy to identify. In further regions, these codes can be re-used. The SC-planning of UMTS networks is similar to the frequency planning of GSM-networks. WFMN7_I_B2, pp

3 129 Proceedings of WFMN7, Chemnitz, Germany Only Macrocells in System Macrocells and Microcell in System Ec/Io (db) Fig. 5. Interference situation when using a micro cell within the system. If a micro cell is added to such a system it dominates within the entire area of investigation as the Best Server providing relative stability and achieving clear-cut boundaries of Soft- Handover zones. Scrambling Code Only Macrocells in System Macrocells and Microcell in System Fig. 6. Best Server in the area of investigation with and without a microcell (Microcell here with SC 277). Case study: Optimizing the UMTS network within the city center of Nuremberg As mentioned above, the planning of UMTS networks relies on sector-macrocells. One possibility of optimization is the electrical or rather mechanical tilt [8] of antennae respectively. In practise, electrical down tilt between 5 and 1 is deployed depending on the type and location of antennae (more in urban center as in rural areas). If the boundary conditions 5 are fulfilled - especially with respect to the location of the microcell -, identifying regions where the microcells can be located at the aim of increasing the performance of the network is an easy talks (Fig. 7, white marked areas). 5 Boundary conditions: Hotspots and Blackspots. Fig. 7. Identified areas for the use of micro cells in the city center of Nuremberg. The common means of optimizing a network, using electrical down tilt antennae, is not sufficient in such areas. By raising antennae the Soft Handover Area may be increased, and by lowering them gaps may occur. Each additionally implemented macrocell in an already dense network can cause further instability. Using the Optimization Triangle for Mobile Networks (Fig. 8) as an example, the results of UMTS networks using microcells shall be demonstrated. Coverage: Power: -85 dbm -6 dbm Capacity: Throughput 25 kbps 32kbps Fig. 8. Optimization triangle for mobile network. The dashed triangle: The network has not been optimized. The values show the average of the measured parameters in the identified areas. The drown triangle: Possible optimum by using microcells. The values show the average of the measured parameters. IV. THE USE OF MICROCELLS IN ORDER TO ACHIEVE CAPACITY EXTENSIONS IN UMTS NETWORKS A. Performing the experiment Quality: SNR -8 db -4 db In Europe, UMTS networks use the Wide Band Code Division Multiple Access (W-CDMA) as access method [6]. The capacity of a cell (macro- or micro) is defined by the bandwidth available. This experiment examines whether a microcell can enlarge the capacity. A comparison of two states regarding throughput and interference situation will be shown. WFMN7_I_B2, pp

4 13 Proceedings of WFMN7, Chemnitz, Germany Present-state: The area of investigation is covered only by a macrocell Combined-state: The area of investigation is covered simultaneously by a macro- and microcell 6 For the experiment, a hotspot is simulated in the area of investigation. Ten test participants (TP) are within close range of the measuring equipment (ME). The TPs are placed in the area as depicted in Fig.9. This area would be the coverage area of an idle microcell. Fig.9. Positions of the test participants in the area provided by the micro cell when it was in the idle-state. First, the measuring equipment establishes a connection to the server and remains within the network throughout the duration of the experiment. The test participants enter the network in groups of two, one group after the other. The chronological order is important, as it enables to check how many participants can be embraced in the network and how many are blocked. The TPs all use the same service, which requires high data rate. A convenient service for this experiment is Mobile TV since it requires a throughput of approx. 114 kbps (payload). B. Evaluation TP1 TP2 TP5 TP6 TP 7 TP 8 TP3 TP4 TP9 TP 1 Macrocell Microcell As expected, the capacity is divided among the participants according to the service required and the resources available at that time. Since all participants use the same service, except for the measuring equipment (downloading continuously 1MB files), one can assume that the capacity should be divided in equal shares. Fig.1 shows the throughput from the measuring equipment in the present-state throughout the duration of the experiment. During the first interval, the total capacity is divided between two TPs and the ME. The throughput measured by ME is approx. 384 kbps. During this interval, the TPs describe the quality of the picture on the screen as very good (Fig.11), which is a sign that the required data rate has been achieved. Within 3 seconds (timeslot 2), the next two TPs start their Mobile TV service. Therefore, the capacity is shared by four TPs and the ME. During the interval of 3 to 6 seconds, the throughput of the ME is reduced to a third. All four participants classify the quality of the picture as good (Fig.11). The TPs get the required data rate for this service in this interval, too Timeslot 1: s-3s, Timeslot 2: 3s-6s, Timeslot 3: 6s-9s, Timeslot 4: 9s-12s Fig.1. Throughput of present-state when increasing the number of participants. At the next logon times (timeslot 3 and 4, Fig.1) the next pair of TPs establish Mobile TV connections. Now, the ME and six, and respectively eight, TPs are logged on to the network; finally, the total capacity is distributed among eight participants. The throughput for the ME does not change. According to the TPs there are drastic changes. In this interval, the picture quality deteriorates, in some cases buffering of data occurs or the connection is lost (Fig.11). In interval 5, all TPs release the connections and the ME regains a throughput of approximately 384kbps Test partitipant 1 Test partitipant 3 2 TP 4 TP 6 TP 8 TP 1 TP Test partitipants in System Subjective notion of TP 1: very good 2: good 3: medium 4: bad 5: buffering 6: interruption Fig.11. Picture Quality from the point of view of the TPs in the present state. 6 For this experiment, the used microcell has a omnidirectional pattern and transmits with a total power of 2 Watt. WFMN7_I_B2, pp

5 131 Proceedings of WFMN7, Chemnitz, Germany Now the following question arises: How does the interference situation change when more participants log on to the system? Ec/Io (db) Fig.12. Interference situation at present-state with growing number of participants Fig.12 shows the increase of interference when more participants establish connections (around 1dB per 2TPs). When eight to ten TPs are served by the same cell at the same time, a deterioration of approximately 3dB is registered. When comparing the present-state with the combined-state regarding the throughput (Fig.1 and Fig.13), one finds that only the first interval, and therefore the period of time when two TPs use the system, is the same. By adding new TPs, the coverage area of the microcell decreases and the impact of cell breathing [7] becomes visible microcell shrank, are served by the macrocell only shows the subjective evaluation of the picture quality of TP1 (Fig.14). While TP1 (at the beginning of the experiment at the edge of coverage of the microcell) describes the picture quality as very good throughout the entire measuring, TP3 (within close distance to the microcell) suffers from buffering of data and finally connection loss. Due to the increase of the interference, the connection is cut off for TP3 as well as ME at the same time. Buffering of data is a consequence of not receiving sufficient data rate. Since the connection of the ME to the network has been cut off, there are no measurements recorded for the last interval (Fig.13). The timer for the required Ec/Iovalue of -12 db is overtaken. This leads to a hard handover with a switch to the GSM network as a consequence Test partitipant 1 Test partitipant 2 2 TP 4 TP 6 TP 8 TP 1 TP Test partitipants in system Subjective notion of TP: 1: very good 2: good 3: medium 4: bad 5: buffering 6: interruption Fig.14. Picture quality described by the TPs during the combined-state. Concerning the micro cell we can conclude the following: The cell shrinks quickly as the load increases. The TPs served by the microcell do not get the required data rate for the service are using. When using the microcell, the interference rises to the extent that the connection is cut off and/or buffering occurs Timeslot 1: s-3s, Chaos area > 3 s (shaded area) Fig. 13. Throughput in combined-state with growing number of system users. The instability of the throughput in the shaded region - chaos-area (Fig.13) can be explained as follows: As the microcell s coverage shrinks, TP1 and TP2 are left out of the coverage of the microcell and are only served by the macrocell. This means that the capacity available increases until the next TPs log on. This behaviour will probably be continued. The fact that at least the first two TPs, after the Fig.15. Throughput in the area of the macrocell when microcell is working to full capacity (macrocell is relieved, possible throughput is approximately 384kbps). WFMN7_I_B2, pp

6 132 Proceedings of WFMN7, Chemnitz, Germany With regard to the macrocell, however, we can assert that it is unburdened when a microcell is used. Therefore, we can surely assume that the microcell has the potential for capacity extension. The subjective evaluation of the picture quality of TP1 reveals that the reference macrocell is unburdened. This assumption has been proved by the measuring results which have been carried out in the coverage area of the macrocell (Fig.15). V. CONCLUSION The performance of UMTS networks when using a micro cell has been investigated. It became clear that the distance between the micro- and the reference macrocell plays an important role. The bigger the distance, the smaller the impact of macro- on microcell and vice versa. Based on this finding, a strategy for optimizing urban UMTS networks by deploying microcells has been demonstrated. The last section of the paper pointed out that the microcell has potential for capacity extensions and can therefore heavily unburden the macro cell. REFERENCES [1] Patrick Herhold, Wolfgang Rave, Gerhard Fettweis, Joint Deployment of Macro- and Microcells in UTRAN FDD Networks, Dresden University of Technology, Mannesmann Mobilfunk Chair for Mobile Communications, D-162 Dresden, Germany. [2] Shaline Kishore, Stuat C. Schwartz, Uplink Throughput in a Single- Macrocell/Single Microcell CDMA System, with Application to Data Access Point, IEEE Transactions on Wireless Communications, Mar 25. [3] Jung-Shyr Wu, Jen-Kung Chung: Performance Study for a Micro Cell Hot Spot Embedded in CDMA Macro Cell Systems, IEEE Transactions on vehicular technology, Vol. 48, No.1, January [4] I.Jami, H.Tao, Micro-Cell Planning within Macro-Cells in UMTS: Downlink Analysis, 3G Mobile Communication Technologies, Conference Publikation No. 489, 22. [5] Joseph Shapira, Microcell Engineering in CDMA Cellular Networks, IEEE Transactions on Vehicular Technology, Vol.43, No.4, November [6] Harri Holma, Anti Toskala, WCDMA for UMTS, John Wiley & Sons, Chichester 2. [7] William.C.Y. Lee, Mobile cellular telecommunications systems, McGraw Hill Book Company, [8] Tero Isotalo, Jarno Niemelä and Jukka Lempiäinen, Electrical Antenna Downtilt in UMTS Network, Institute of Communication Engeneering, Tampere University of Technology, P.O.BOX 553, FIN-3311 Tampere, Finland WFMN7_I_B2, pp