You are on page 1of 6

Performance Study on the Effects of Cell-Breathing in WCDMA

Kay Leong Thng1,2, Boon Sain Yeo1 and Yong Huat Chew1
2

Institute for Infocomm Research, 21 Heng Mui Keng Terrace, Singapore 119613 Electrical and Computer Engineering Dept., National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260 breathing may even give an advantage with the possibility of improving the capacity in some cells with higher traffic by redirecting MSs to surrounding cells. However, this deployment is not quite realistic, and cell-breathing does not provide much advantage from a system capacity point of view if the traffic demands in all the cells are high. As shown in [7], there is a trade-off between capacity and coverage. It is well known that the coverage of a cell has an inverse relationship with the density of MS in a cell. However, the exact relationship between capacity and coverage is still very much indistinct presently. With the rapid development of 3G mobile communication system and the possibility that many countries will be adopting this system in the next few years, it is crucial that tools are made available for network planners. Simulators are often the tools being used. In this paper, a detailed framework of a 3G system simulator, based on 3GPP, will be developed. With the development of this simulator, various characteristics of a 3G system can then be studied intensively (e.g. cell breathing, biterror-rate and call blocking). Besides, work on the optimization of the system or network planning can also be carried out with the aid of this tool. This paper continues in Section II with an overview on the structure of the simulator which has been developed for our study of cell-breathing. Section III describes the simulation scenario with one cell and the corresponding result obtained is presented and discussed. Section IV extends the simulation to two cells. Finally, some conclusions are given in Section V.

Abstract The performance of CDMA cellular systems over the air is generally interference-limited. A phenomena arising from this is the effect of cell-breathing in 3G cellular systems. Cellbreathing is the expansion or contraction of the effective coverage of a cell in response to the number of active mobiles (MSs) in a network. If it is not well controlled, communication failure may result. With 3G offering different classes of services, planning and dimensioning of such high-quality radio networks thus requires extensive planning tools. Despite the progressive rolling out of 3G systems worldwide, the effects of cell-breathing are not well understood. In this work, a simulator is set up to study the effect of cell breathing. Simulation results show how the area where reliable coverage can be provided by a base station (BS) may varies as the density of MSs changes within it. Current work emphasizes on the downlink simulation with a single BS followed by a similar simulation involving 2 BS. The results demonstrate how cell-breathing may cause communications failure.

I. INTRODUCTION
Universal Mobile Telecommunication System (UMTS) is a third generation mobile network designed for multimedia communication which enables person-to-person communication with high quality images and video. With UMTS, access to information and service will be enhanced by higher data rates, and Wideband CDMA (WCDMA) is to be employed over the air interface. With 3G mobile network based on WCDMA, users (or MSs) can occupy the entire allocated frequency and time domain, and are distinguished from each other through the use of unique codes. The codes for other MSs will appear as noise, in which either codes are non-orthogonal or there is loss of orthogonality due to the presence of multipath. As the number of MSs increases, the level of interference rises. From [1], to successfully demodulate the transmitted user data bits, the SIR must be greater than a threshold. Failure to do so will result in momentarily poor link quality and even the call will be dropped. The greater the number of users in a cell, the greater the interference, hence, the larger will be the required received power. If we consider that a MS and a BS have a fixed maximum transmit power, changing the required received power will have the same effect as changing the maximum path loss that can be tolerated between the BS and the MS. In other words, the effective area where reliable quality of service (QoS) can be provided by a BS will decrease. This is known as cell-breathing. Thus, the fundamental drawback of cellbreathing is that the desired SIR of a MS may not be achievable at some times of the day even though he may be at the same location, and that experience will be truly annoying. A number of researches have been carried out to reduce this impact of cell breathing [2-5]. In [6], it is shown that cell-

II. SIMULATOR OVERVIEW


Accuracy of a simulator depends very much on the choices made for the physical layer model. By and large, a compromise has to be made between (i) accuracy and (ii) complexity when developing such simulator. Based on the work from [10], there are basically 2 classes of simulator Static & Dynamic. Static Simulator are based on Monte Carlo approaches and essentially work by dropping MSs in a defined network layout, and thereafter using some algorithm to decide which proportion of the MSs can be correctly served. Then the process is repeated with a number of drops, where in each case the MS spatial distribution and numbers correspond to a realization of a global statistical model of network load. The performance indicators will eventually converge, after which the process can be re-run for another set of parameters. On the other hand, a Dynamic Simulator works on the principle of two time scales namely short-term and long-term. If interest is on admission control, overload control, handover and call dropping, then there is no requirement to consider

0-7803-9206-X/05/$20.00 2005 IEEE

micro time scales, and are referred to as Long-Term Dynamic Simulator. If interest is focused on functions such as packet scheduling, fast cell selection, random access channel or forward access channel performance, then the models should include short timescales and this kind of modeling is referred to as Short-Term Dynamic Simulator. In that paper, it has been justified that a dynamic simulator has an edge over static simulator in terms of accuracy. It is because a static simulator has no concept of time unlike a dynamic simulator which is able to take into consideration characteristics of the system like MSs mobility, admission control, handover, call generation and generally timedependent algorithms. A short-term dynamic simulator also takes into account the detailed model of the effects such as channel fading, inner loop power control, packet scheduling and fast cell selection etc. However, these characteristics are not considered in a longterm simulator. Thus, it can be concluded that short-term dynamic simulator offers the greatest flexibility and potentially a better accuracy than the long-term one, albeit at the expense of extra-complexity at the system level. With a long term goal in mind, our aim is to develop a simulator with the concepts of a short-term dynamic simulator. The characteristic of a short-term dynamic simulator can be summarized as follows: Taking the time slot duration as the time step of the system. For example, if the inner power control loop has time duration of 0.666ms, we shall have the simulator to work on time steps of 0.666ms, i.e., the calculation of SIR values are done and updated every 0.666ms. This example provided us with an intuitive approach to model our simulator algorithm. In general, the design of this simulator is structural, with a top-down approach. The exact algorithm of our simulator will be discussed more in details over in this section. The simulator will be built upon 2 main modules Environmental Module and Protocol Module. Contained in these 2 modules will be sub-models, which can be considered as C-functions that the main program will call when the simulation is run. IIA Environmental Module The Environmental Module serves as an interface between the user of the simulator and the simulator itself. In will contain an interface for the reading of input parameters or data and another interface where the results of simulation are shown. The structure of this module is shown in Fig. 1.
Parameter

Parameter Structure and Radio Channel This consists of 3 data structures namely MobileStation, BaseStation and System. MobileStation and BaseStation carry variable parameters of all MS and BS respectively, while System carries the variable parameters for radio channel behaviours such as the 3G operating parameters. Systems will also carry the simulation results. The appendix shows the parameters to be carried by System. Simulate 3G network This contains the whole of the simulation program and is part of the Protocol Module. Most of the parameters used in this module is based on release 99. Results This provides the interface which presents the results to the MS. II.B Protocol Module The Protocol Module contains the heart of the simulation program. It consists of 4 main functions namely Generate, Propagate, Receive and Control. These functions in turn contain numerous sub-functions. Fig. 2 shows the Behavioural Module.
Simulate 3G network Generate Propagate Receive Control

Fig. 2 Behavioural module In general, the sequence of the program shall be top-down starting (after Initialize) from Generate to Propagate, Receive and finally Control. Each main function will be started with a Trigger (a call) from the main controller. Upon completion of its task, it will response to the main controller with a status of Ready (a return). Similarly, the sub-functions within each main function will be started and completed in the same way. Fig. 3 shows the highest level data flow diagram of the protocol module and shall explain the design clearer. It shows the flow of data after Initialize. A description of the functions and its sub-function, in the sequence of the flow of the simulation are as follows:
Parameter Structure

R
1. Initialize

T
6.
Control Simulation

Data flow Signal flow

Radio Channel

R
3.
Propagate

2. Generate

R
4. Receive

T
5. Control

Simulate 3G network

Results

Radio

Fig. 1 Environmental module

Parameter Structure

Results

Fig. 3 Data flow diagram of the overall protocol module

Initialize This module marks the start of the simulation program. Characteristics of all MS and BS are defined. The attributes of each MS (location, speed of motion, direction of motion, service class and duration of call) are generated randomly. The required connection and transmission power for each entity are also established with minimum transmission power. Generate In this state, existing calls will first be checked if it is to be ended normally i.e. the call duration has expired. New calls and its duration will be then generated according to a certain statistical distribution. 3GPP has stated 4 service classes namely - conversational real time, interactive, streaming, background. We can represent each of these classes with a fixed data rate. The data rate corresponding to each class of service and an example of its application are shown in Table 1.
QoS classes Conversational real time Interactive Background Streaming Data speed 12.2kbps 64kbps 144kbps 384kbps TABLE 1 Error-tolerant Voice, video Voice messaging Fax Audio, video Error-Intolerant Interactive games, Telnet Web browsing, ATM Email FTP data transfer

+ (44.9 6.55 log hb ) log R + C m

L = 46.3 + 33.9 log f 13.82 log hb a(hm )

(1)

where f is 2150MHz in downlink and 1950MHz in uplink, hb = 8m is the effective height of the BS, hm =1m is the height of MS antenna, R is the distance between BS and MS in km, a(h m ) = (1.1 log f 0.7 )hm (1.56 log f 0.8) and C m is 0dB for medium-sized city and suburban centers.

Received Power Calculation In this sub-module, the power at the receiving end is determined after propagating through the channel. The result of its computation will be used by sub-function Interference for the determination of SIR.
Receive This state consists of 2 sub-functions namely Interference Calculation and SIR Calculation. With the determination of interferences in the uplink as well as downlink, the SIR in the uplink and downlink can then be calculated. From [5] and [6], a detailed approach to determine the interference and SIR has been established according to the following equations.

Uplink SIR
d k ,l = S k k ,l x k ,l W N 0

Table 1. Data rate corresponding to each class of service. From [2, 3, 11] the call arrival distribution, duration, and average speed of each MS for each class of services are tabulated in Table 2. Probability of generating each new call is based on a Poisson distribution; duration is based on a exponential distribution; speed is based on Gaussian distribution.
Data speed 12.2kbps 64kbps 144kbps 384kbps TABLE 2 Arrival rate/min 0.03333 0.01667 0.00833 0.00833 Duration 120s 288s 192s 300s Avg Speed 50km/h 40km/h 30km/h 20km/h

( ) +
L l i =1

x d v S j j j ,l j ,i j =1 j k
K

W Rk

(2)

Downlink SIR k =
W N 0

( )

S d k ,l k x k ,l d + 1 j vjS j j ,l x j ,l + v j S j d j ,i x j ,i + Tl d k ,l
j =1 i =1 j =1 i l K

W Rk

Table 2. Statistical call distributions, call duration. Propagate This state consists of 3 sub-functions namely Mobility, Signal Propagation and Received Power Calculation. Mobility In this sub-function, the new position of each MS will be calculated according to its old position, velocity, simulation time step and predefined route. Signal Propagation In general, this sub-function will determine the drop in transmitted power along the channel (both uplink and downlink) given the transmission power. In this sub-function, the propagation model is based on COST-Hata-Model. The loss in dB is given by:

(3) In (2) and (3), L is the total number of cells and K is the total number of MSs in the area under consideration. The notation x j ,i is used as an assignment variable so that x j ,i = 1 if MS j is connected to BS i, or else x j ,i = 0 . The index l is used to represent the reference cells where the kth MS is attached to, i.e., x k ,l = 1 and the assignment of BS is based on largest received power. Rk is the data rate of k (bps), W is the chip rate (Hz) and the value used is 3.84Mcps, Sk is transmission power (dBm) assigned to kth MS, k is the desired received E b / N 0 of kth MS (downlink) or BS (uplink) and dk,l is the attenuation between BS l and MS k (dB). k is the orthogonality factor in the downlink of k ranged between [0..1] and a value of 0.75 is used for those MSs in cell l in our simulation. Similarly vk is the activity factor of and ranged between [0..1] and the value used is 1.0. Tl is the transmission power at BS l used by all the common channels and the value used is 36.0dBm. There are power controls in both the uplink and downlink associates with maximum allowable transmit power. (N0)l is the thermal noise power density at l given by -169dBm/Hz. For any parameter y

. Although it is not clearly given in decibel, its linear value is y distinguish, some of these parameters such as activity factor, path loss, etc., should be different in the uplink and downlink in practical scenarios. However, in our simulation, we assume that these parameters are the same.
Control This state consists of 4 sub-modules namely Admission, Quality, Handover and Power Control.

will be made to increase the transmission power provided the transmitting entity is not yet transmitting at full power. Similarly, in the event the SIR is above the target SIR, power control will step down the transmission power of the transmitting entity.

III. SIMULATION WITH ONE CELL


The program that developed is next used to illustrate the cell breathing effect. For the studies of cell-breathing effect, RPCAC will only be active during the transient state of the simulation. This is to ensure that steady state has been reached before any simulation results are taken. RPCAC will make inactive upon the system reaching stability and when the location and arrival of MS is to be manipulated to show cellbreathing, i.e., in other words, all newly generated calls are allowed into the system. For simplicity, we also consider only single class (conversational real time) service.
Scenario In this simulation, the coverage area recorded is specific on the downlink for call-type of 12.2kbps. A single cell is located at the centre of the simulation area, which is a 100 100 grid square. Each grid division corresponds to a distance of 120m. MS and its class of service are distributed randomly throughout the whole simulation area. The total number of MS involved in the test is 500. The purpose of having this large number of MS which a single cell certainly cannot handle is to ensure that there is a definite possibility of overloading the network such that cell-breathing can take place. As the average call duration for voice calls is estimated to be 120s, we assume stability of the system to be reached by then. Upon the system reaching stability, all MS will move towards the BS at a speed of 12m/s. Cell-breathing can then be observed and analyzed. Here, coverage is defined as the maximum distance from a BS where SIR is above the required threshold and hence reliable service can be provided. Table 4 below shows a summary of the simulation condition.
TABLE 4 Parameters No of BS No of MS Simulation time step Map scale No of simulation steps to stability Convergence rate of MS Values 1 500 10ms per step 120m per division 12000 steps 12m/s per step

Admission This sub-function will decide upon whether to let a new call into the system after successful synchronization. Call admission algorithm is not specified by 3GPP and may vary with different operators. In our simulator, the call admission control implemented is based on Receive Power Call Admission Control (RPCAC). In general, a new call is allowed if the total received power at the BS, l, is below a threshold, Ith. When a new MS requests a call, the algorithm will estimate the increase in I, I, that the MS will cause to the system. The call will be admitted if the overall system load is still below the threshold value, otherwise it will be rejected. I + I < Ith

(4)

Quality This sub-function will estimate the received signal strength and drops a call on detected quality failure. Here, quality estimation is based on the SIR calculated. If the SIR stays below a threshold for a preset time, then it will be deemed quality failure and the call will be dropped. Table 3 shows the threshold values for the 4 classes of service [7].
Data speed 12.2kbps 64kbps 144kbps 384kbps TABLE 3 Uplink Threshold 6.1 4.0 3.1 3.0 Downlink Threshold 7.9 4.0 4.5 4.5

Table 3. SIR Threshold for different class of service. Handover Soft Handover (SHO) is implemented in our simulator. In this scheme, the network measures the uplink connection(s) and receives measurement results from downlink connection(s) made by the UE. SIR (both uplink and downlink) will be calculated at state Receive in every simulation step. A handover to another BS will occur when the SIR (both uplink and downlink) exceeds that of the serving BS by a certain handover threshold. This handover threshold is stated in [7] as 3.0dB. A function call to verify admission will occur to check if handover to the new BS is successful. Power Control This sub-function uses the instantaneous SIR derived in Receive to determine if there if a need for increasing transmission power. A comparison is made between the measured SIR and the target SIR for the different classes of service. If the SIR measured is below the target SIR, an attempt

Table 4. Summary of simulation conditions. Results and discussion Fig. 4 shows the graphical plots of coverage distance with simulation time steps. Upon reaching steady state after the start of the simulation, the radius where reliable signals can be provided by the BS is approximately 3100m. It can be seen that with the onset of convergence of MS towards the BS, there is generally a decline in the radius of reliable coverage. A drastic decline in the reliability is observed when all MS have moved within 1800m (each simulation time step in the figure is 20ms)

from the BS. This sharp drop in radius shows that the system requires a significant number of MS, active at the same time, to trigger off cell-breathing which in this case is a great drop in area where reliable signals can be provided. The resulting radius where reliable signals can be provided by the BS is then approximately 2775m, which corresponds to a 10% decrease in radius.

TABLE 6 Parameters No of BS No of MS Simulation time step Map scale No of simulation steps to stability Convergence rate of MS Values 2 500 10ms per step 120m per division 12000 steps 12m/s per step

Table 6. Summary of simulation conditions. Results and discussion Fig. 5 shows the combined graphical plots for BS A and BS B with respect to time. With the concentration of MS shifting from BS A to BS B, the area where reliable coverage can be provided by BS A and B are seen to be reciprocal of one another, which coincides with theoretical analysis for the characteristics of cell-breathing. The magnitude of expansion or contraction of the 2 BS coverage areas is in the same degree as with the case for 1 cell. This result provides an insight that cell-breathing is not infinite. Cell-breathing is in general accounted by the spatial distribution of MSs. The cap to the maximum transmission power allowed for a transmitter is also another factor. This limit on transmission power causes MS that are not receiving the target SIR to be dropped early and thus, interference within the system are reduced before it gets too large to trigger off further cell-breathing. An important observation from this simulation is the possibility of uncovered area resulting from cell-breathing. This period of unreliability is observed when the MS starts moving towards BS B (between simulation time step 25000 to 38000 as shown in Fig. 6). It is during this period of time that the shrinkage of reliable coverage by BS B may not be well compensated with a corresponding expansion of reliable coverage by BS A. Thus, giving rise to potential communication failure.

Fig. 4 Coverage plot for 1 cell.

Table 5 contains a summary of the total number of calls generated, blocked or successful.
TABLE 5 12.2kbps 1851 1311 70.9% 307 16.6% Call Type 64kbps 144kbps 989 543 918 524 92.8% 96.5% 35 16 3.54% 2.94% 384kbps 576 563 97.7% 5 0.87%

Generated Dropped % Dropped calls. Success % Successful calls.

Table 5. Summary of simulation results.

The low percentage of successful calls by MS stems from the fact that there is only 1 BS supporting such a large number of MS. One would not expect to see this in practice because we purposely overload the system to observe the cell breathing impact. In practice, proper cell planning would require more BS to support the number of MS and service requests.

IV. SIMULATION WITH TWO CELLS


Scenario In this simulation, 2 cells are located across the simulation area, which is a 100 100 map square. BS A is located at coordinate point (35, 35) while BS B is located at (65, 65). MS and its different classes of service are distributed close to BS A at the start of simulation. Upon the system reaching stability, all MS will move towards BS B. Cell-breathing can then be observed and analyzed. In this simulation, the coverage area recorded is specific on the downlink for call-type of 12.2kbps. Table 6 below shows a summary of the simulation conditions.

Fig. 5 Coverage plot for BS A and BS B.

[9] TABLE 7 12.2kbps 931 697 74.9% 155 16.6% Call Type 64kbps 144kbps 544 322 482 306 88.6% 95.0% 36 14 6.62% 4.35% 384kbps 328 317 96.6% 11 3.35%

Generated Dropped % Dropped calls. Success % Successful calls.

Table 7. Summary of simulation results.

F. Babich, L. Deotto, A Formal Approach to Modeling and Performance Analysis of Shared Channels for Real-Time Services in W-CDMA 3G Systems, 52th IEEE Vehicular Technology Conference, vol. 4, 24-28 Sept. 2000, pp. 1639 1645. [10] E. Villier, L. Lopes, S. Lambotharan, Approaches to Modelling the Physical Layer Performance in a UMTS Radio System Simulator, 3rd International Conference on 3G Mobile Communication Technologies, Conf. Publ. No. 489, 8-10 May 2002, pp. 560 564. [11] S.T. Yang and A. Ephremides, Resolving the CDMA Cell Breathing Effect and Near-Far Unfair Access Problem by Bandwidth-space Partitioning, 53rd IEEE Vehicular Technology Conference, vol. 2, 6-9 May 2001, pp. 1037 1041.

V. CONCLUSIONS
In this paper, the structure and features of a 3G simulator were proposed and discussed. Simulations were then carried out to study the effect of cell-breathing via the variations of MS density, firstly with a single cell, followed by 2 cells. The result obtained shows that with the increase in density of MS, the coverage area from a cell will generally decrease. This is consistence with theoretical characterization of cellbreathing. A more concerned effect of cell-breathing is the possibility of communication failure as a result. Thus, to provide a good and reliable communication system, network planning needs to be meticulous in ensuring a proper control on the extent of cell-breathing. Another interesting result is that cell-breathing is finite. In other words, one can neither expect the coverage of a cell that is subjected to a high density of MS to decrease to zero nor can the coverage of a cell with a low density of MS be extremely large. The result presented is important in providing an insight into how coverage area, on the downlink, is also being affected with a restrain on transmission power. Further studies are to look into how to jointly consider cell breathing effect in cell planning, especially in heavy load conditions, so that the network can maintain at a reasonably low dropped call probability.
REFERENCES
[1] [2] P R Gould, Radio Planning of Third Generation Networks in Urban Areas, 3rd International Conference on 3G Mobile Communication Technologies, 2002, Conf. Publ. No. 489, 8-10 May 2002, pp. 64 68. S.T. Yang and A. Ephremides, Resolving the CDMA Cell Breathing Effect and Near-Far Unfair Access Problem by Bandwidth-Space Partitioning, 53th IEEE Vehicular Technology Conference, vol. 2, 6-9 May 2001, pp. 1037 1041. A.I. Zreikat and Al Begain K., Soft Handover-based CAC in UMTS System, 10th International Conference on Telecommunications, vol. 2, 23 Feb-1 March 2003, pp. 1307 1312. J. Yang and J.S. Lin, Optimization of Power Management in a CDMA radio Network, 52nd IEEE Vehicular Technology Conference, vol. 6, 24-28 Sept. 2000, pp. 2642 2647. M.MAl. Akaidi and H. Ali, Performance Analysis of Antenna Sectorisation in Cell Breathing, 4th International Conference on 3G Mobile Communication Technologies, Conf. Publ. No. 494, 25-27 June 2003, pp.98 103. A. Jalali, On Cell Breathing in CDMA Networks, IEEE International Conference on Communications, vol. 2, 7-11 June 1998, pp. 985 988. V.V. Veeravalli and A. Sendonaris, The Coverage-Capacity Tradeoff in Cellular CDMA Systems, IEEE Transactions on Vehicular Technology, vol. 48, no. 5, Sept. 1999, pp. 1443 1450. J. Buczynski, P. Gajewski and J. Krygier, Modelling of the Third Generation Mobile System, IEEE AFRICON, 1999 vol. 1, 28 Sept.-1 Oct. 1999, pp. 251 256.

Appendix List of parameters and its values used by Systems


Symbol chip_rate step_time N_THERMAL_NOISE ORTHO_FACTOR Ith COMMON_CH_PW HO_DL_THRESHOLD HO_UL_THRESHOLD UP_POWER_STEPSIZE DOWN_POWER_STEPSI ZE FREQ_DL FREQ_UL EFF_HT_BS EFF_HT_MS MS_TX_MAX MS_TX_MIN BS_TX_MAX BS_ TX_MIN BS_max_ power_service Description Chip rate of system Each simulation step time Thermal noise power density Orthorgonal factor Total received power at BS for call admission. Common Channel Power Handover DL threshold Handover UL threshold Up powercontrol stepsize Down powercontrol stepsize Frequency of DL transmission Frequency of UL transmission Effective height of BS. Effective height of MS. Max transmission power of MS. Min transmission power of MS. Max total transmission power of BS. Min total transmission power of BS. Max power per service for 12.2kbps, 64kbps, 144kbps and 384kbps. SIR TARGET FOR VARIOUS SERVICES Uplink / Downlink Uplink / Downlink Uplink / Downlink Uplink / Downlink Data type/Remarks 3.84MCPS 10ms -169dBm. 0.75 100dB 30dBm 0.5dB 0.5dB 1dB 1dB 2150MHz 1950MHz 30m 2m +30dBm +5dBm +43dBm +5dBm. +30dBm, +36dBm, +36dBm, +38dBm respectively +6.5dB / +4.5dB +4.0dB / +4.0dB +3.5dB / +1.5dB +3.0dB / +1.0dB

[3] [4] [5]

12.2kbps 64kbps 144kbps 384kbps

[6] [7] [8]

You might also like