GSM Frequency Planning with Band Segregation for the Broadcast Channel Carriers

Transcription

1 GSM Frequency Planning with Band Segregation for the Broadcast Channel Carriers F. Galliano (1), N.P. Magnani (1), G. Minerva (1), A. Rolando (2), P. Zanini (3) (1) CSELT - Via G. Reiss Romoli, Torino (Italy) (2) formerly with CSELT, now with Mercury One2One (3) formerly with CSELT, now with Alcatel Abstract In this paper we investigate the performance of fixed frequency planning of the GSM system considering dedicated bands for TCH and BCCH carriers respectively (the BCCH carriers are the ones which host the BCCH channel). This analysis aims at assessing whether this frequency planning strategy (which in the following will be indicated as frequency planning with band segregation ) brings to better system performance with respect to the usual frequency planning strategies, i.e. common band for TCH and BCCH carriers. A number of representative deployment scenarios are investigated. 1. Introduction Optimizing the use of the available frequency resources is fundamental in order to increase the overall capacity of a GSM system. To achieve this, a number of solutions may be adopted, like improved radio resource management policies, deployment of advanced radio features, introduction of advanced resource planning strategies such as Dynamic Channel Allocation or Fractional Loading. This paper addresses a fixed frequency planning strategy, based on the use of dedicated bands for TCH and BCCH carriers respectively. References to this technique can be found in [1] and [2]. In the following, frequency planning with band segregation is compared to conventional fixed frequency planning with common TCH/BCCH band. Three cases have been considered: 1] system performance analysis considering an ideal scenario; 2] development of frequency plans with and without band segregation with reference to a realistic cell layout; 3] system performance analysis of a realistic scenario taking into account in field cell dimensioning and parameters. The simulation results reported in case 2 were obtained with the frequency planning tool FREQUENT (FREQUency assignment) developed in CSELT which takes into account the mutual average interference between cells evaluated on a priori basis (e.g., without considering system functionalities nor users behavior); these results are therefore relevant for the planning process. On the contrary, results presented in case 1 and 3 were obtained by means of a software tool developed in CSELT (TOTO - TDMA Oriented software tool [3]), which models the main functionalities of the GSM system and takes into account both propagation data, system functionalities (e.g. power control, DTX), and users' behavior; these results are therefore relevant for the validation of the frequency plan. The two sets of results, although referring to two different steps of system deployment and therefore being obtained with different methodologies, show the same trend. 2. System performance analysis considering an ideal scenario 2.1 Simulation scenario and considered functionalities Simulations were carried out considering an ideal scenario (i.e., regular layout of hexagonal cells, signal level distribution according to Hata propagation model), with the same number of transceivers for each cell. Frequency planning is performed according to a fixed assignment criterion, where frequencies are distributed according to a regular cluster basis. A pool of 48 carriers was assumed. A few cases were considered, where the number of carriers per cell (N) was ranging from 4 to 7, depending on the cluster size. For the frequency planning with common band for all carriers (i.e. with TCH and BCCH sharing the same band), frequencies were assigned to cells according to cluster size 48/N; the BCCH carriers were assigned to the cells assuming all 48 carries available, in such a way that co-channel cells belonging to geographically adjacent clusters have different BCCH carriers. In the case of frequency planning with band segregation, carriers were reserved for exclusive BCCH use throughout the network (i.e., BCCH carriers use a cluster size ), and TCH carriers used the remaining frequencies (36) with cluster size (36)/(N-1). Table 1 summarizes the considered simulation scenarios. The simulated network was considered to be synchronous, with low mobility user terminals, and a traffic load corresponding to an average 2% blocking probability. Discontinuous transmission (DTX),

2 handover, Power Control (PC, only quality based case was considered) were taken into account, based on GSM Specification; adjacent channel interference and frequency hopping were not taken into account. 2.2 Simulation results Results obtained for the downlink are summarized in figure 1, which shows the values at 1% of the C/I cumulative distribution versus the number of transceivers per cell. Two cases are considered, corresponding to active PC or PC and DTX. The frequency plan with band segregation provides better overall performance for cell carrier equipment up to 6 transceivers per cell (corresponding to a reuse 7 for TCH) when both PC and DTX are active. Actually, this is the result of two opposite effects. As figure 2 shows, the performance of the BCCH carriers worsen when moving from the case without band segregation to the case with band segregation, due to the shorter reuse distance (with band segregation, only carriers are dedicated to BCCH whereas without band segregation all 48 carriers are available for BCCH planning). On the contrary, figure 3 reveals that TCH performance improve substantially with band segregation. In this case, TCH carriers are not affected by the interference coming from the BCCH carriers (which, according to the GSM Specifications, are always transmitting with the maximum allowed power). This also allows to maximize the benefits of system functions such as power control. The introduction of DTX further improves TCH performance. However, if the number of transceivers per cell increases, thus making shorter the reuse distance of the TCH carriers, the performance improvement gradually disappears. In figure 1 it can be noted that the results with 4 TRX/cell are not aligned with other cases, in fact the performance improvement achievable with dedicated bands is slightly worse than the one achievable with 5 TRX/cell. This is due to the fact that with 4 TRX/cell the quality of the BCCH carriers is substantially lower than the quality of TCH carriers (see figure 2 and 3); this implies that almost all values in the lower part of the cumulative distribution (from which the 1% value is extracted) belong to the BCCH carriers, whose relative weight is therefore definitely higher than in other cases. If we referred to the 2% values of the C/I cumulative distributions (rather than 1%), this phenomena disappears. Figure 3 highlights that the TCH quality decreases more rapidly with band segregation when the reuse distance decreases; in fact, with the common band approach, TCH carriers suffers from the interference coming from BCCH carriers, which is basically constant, regardless of the number of carriers per cell. Simulation results do not show any advantage of the frequency planning with band segregation for the uplink. In fact, in the uplink, the increased interference due to a smaller reuse distance of both TCH and BCCH carriers is not balanced by any positive effect, since the BCCH carriers in uplink do not behave differently than normal TCH carriers (i.e. PC and DTX can be applied to BCCH carriers as well as to TCH carriers). Common Band Band Segregation TRX/cell # carriers Overall reuse # carriers BCCH reuse TCH reuse (*) (**) (*) Results obtaind by interpolation of cases with 45 and 5 carriers (**) Results obtained by interpolation of cases with 42 and 4 carriers Table 1: TRX assignment scheme C/I DL 1% - Global 1 C/I 1% dif C/I 1% com C/I 1% com DTX C/I 1% dif DTX TRX/cell Figure 1: Downlink C/I@1% (BCCH and TCH carriers)

3 C/I DL 1% - BCCH C/I 1% bcch dif C/I 1% bcch com C/I 1% bcch com DTX 8 C/I 1% bcch dif DTX TRX/cell Figure 2: Downlink C/I@1% (BCCH carriers) C/I DL 1% - TCH C/I 1% tch dif C/I 1% tch com C/I 1% tch com DTX C/I 1% tch dif DTX TRX/cell Figure 3: Downlink C/I@1% (TCH carriers) 3. Development of frequency plans with reference to a realistic cell layout In case 2 we have built three different frequency plans on a realistic set of cells using the software tool FREQUENT developed by CSELT to solve frequency allocation problems. The first one is defined using a common band to plan the BCCH and TCH carriers of the cells, while the second and the last one using two different configurations of band segregation. The total band available for the frequency allocation was considered as constituted by 5 frequencies shared as follows: Frequency plan BCCH band TCH band 1 1o5 1o5 2 1o o5 3 1,3,5,7,,,,, 2,4,6,8,1,,,,,2,,1,21,23,25,27,2 22,24,26,28, 3o5 Table 2: band configurations The frequency plans accomplished by means of FREQUENT use a set of realistic inputs both in terms of interferential description and in terms of reuse constraints set. The description of the interference between the cells of the plan is given by the interference matrix indicating for each couple of cells (i,j) the C/I induced by the cell j on the cell i when both cells use the same frequency (co-channel interference); the interference due to the adjacent channels is estimated evaluating the division between the co-channel C/I and the parameter NFD (Net Filter Discriminator). The reuse constraints set is constituted by:

4 cell constraint, implicating a minimum distance equal to 3 between frequencies assigned to the same cell; site constraint, implicating a minimum distance equal to 2 between frequencies assigned to different cells belonging to the same site; 4 khz adjacencies constraint, implicating a minimum distance equal to 2 between frequencies assigned to 4 khz adjacent cells; 2 khz adjacencies constraint, implicating a minimum distance equal to 1 between frequencies assigned to 2 khz adjacent cells. The area on which the three frequency plans are accomplished is constituted by 622 cells and 2435 carriers, the elaboration time (CPU time) used for the definition of each plan is equal to 1 hours on a Sun Spark Ultra Workstation. The different plans are compared as a function of: the minimum C/I ratio associated to the BCCH carriers set; the minimum C/I ratio associated to the TCH carriers set; the number of carriers under the threshold of db; the C/I quality distribution of the BCCH carriers within the cells; the distribution of the C/I associated to the BCCH carriers; the distribution of the C/I associated to the TCH carriers. The results are shown in the following tables and diagrams: Plan 1 Plan 2 Plan 3 Minimum C/I of the BCCH carriers.8 db 7.41 db 8.62 db Minimum C/I of the TCH carriers.7 db 7. db 7.54 db Number of under threshold carriers 27 2 Table 3: C/I results Plan 1 1% Plan % 27.8% % 6.4% 1.8%.5% Plan % 33.1%.5% 8.8% 2.6%.2% Table 4: BCCH quality distribution within the cells (maximum cell equipment is equal to 6 TRX) BCCH carriers Quality plan 1 plan 2 plan 3 Figure 4: BCCH quality distribution < db - db -2 db 2-25 db 25-3 db 3-35 db > 35 db Plan 1-1.4% 33.3% 32.8% % 5.1% 3.4% Plan 2.3% % 37.6% 28.6% 1.5% 3.% 2.1% Plan 3.2% 2.6% 41.3% 25.5% 7.1% 3.% 1.4% Table 5: C/I distribution of BCCH carriers

5 BCCH carriers < > 35 db plan 1 plan 2 plan 3 Figure 5: C/I distribution of BCCH carriers < db - db -2 db 2-25 db 25-3 db 3-35 db > 35 db Plan 1-2.4% 41% 2.% 5.6% 1.7% 1.4% Plan 2 1.4% 3.7% 37.6% 2.8% 5.7% 2% 1.8% Plan 3 1% 3.6% 38.3% 1% 7% 2.4% 1.7% Table 6: C/I distribution of TCH carriers TCH carriers < > 35 db plan 1 plan 2 plan 3 Figure 6: C/I distribution of TCH carriers As shown in the set of tables and figures the frequency plan with common band seems to guarantee the best results, in particular for the BCCH carriers that are chosen after the planning assigning the BCCH channel to the carrier with the best C/I in the cell. This procedure cannot be performed in case of band segregation because only one frequency assigned to a generic cell is compatible with the BCCH planning; in case of segregation the first configuration guarantees the better results for the broadcast channels. As far as the TCH channels are concerned, the results are similar in case of common and segregated band but in the second and third case these results can become worse if the average number of carriers per cell increases, owing to the most intensive effects of the interferential condition and cell/site/adjacency constraints. 4. System performance analysis of a realistic scenario 4.1 Simulation scenario and considered functionalities In order to evaluate the performance of frequency planning with band segregation under realistic conditions, system simulations were carried out considering the realistic layout of the central area of an Italian city. Propagation data were evaluated with a model validated by means of field measurements [4]. Real cell dimensioning was taken into account (ranging from 2 to 7 carriers per cell, 5 on average). 41 available frequencies were considered; in case of frequency planning with dedicated bands, frequencies were reserved for BCCH. Relevant frequency plans were obtained taking into account all adjacent constraints (see section 3) due to technology by means of the tool FREQUENT. Real measured data traffic per cell were used for the system simulation. The simulated network was considered synchronous, with low

6 mobility user terminals. Discontinuous transmission (DTX), handover, Power Control (PC, only quality based case was considered) were considered, based on GSM Specifications; adjacent channel interference was taken into account. 4.2 Simulation Results Simulations have shown an high load in the network characterized by an overall blocking probability higher than the 2% considered in the ideal case. In this case the overall downlink performance in terms of C/I cumulative distribution do not improve by adopting the frequency planning with band segregation (see figure 7). Anyway, this result is not in contradiction with the one obtained by the analysis of the ideal scenario considered in section 2 above, since different conditions apply. In the realistic scenario studied here, the network load is higher and non-uniformly distributed. Under these conditions, the interference contributions from BCCH and TCH carriers tend to become more balanced; as a consequence, the use of dedicated bands do not lead to a more efficient use of functionalities such as power control and DTX. Moreover, in the scenario considered here, the number of transceivers per cell is quite high (up to 7, 5 on average) and therefore the reuse factor is quite low. Finally, the frequency planning considered for this realistic scenario was evaluated taking into account all adjacent constraints due to technology, which were not considered in the ideal case (where the cluster approach was used) and which effectively reduce the degrees of freedom available for the planning algorithm (in particular, when the number of available carriers is low, as in the dedicated bands case) with obvious impact on the final frequency plan. Therefore, in the considered realistic case, the TCH performance in the band segregation case are not capable of balancing the degradation of the BCCH performance. In the uplink, simulation results do not show any advantage of the frequency planning with band segregation, in line with results previously obtained in the ideal case Prob( C/I < x value ) COM SEG C/I 5. Conclusion Figure 7: cumulative C/I distribution (BCCH and TCH carriers) The system performance analysis carried out considering an ideal scenario (section 2) has shown that the frequency planning with band segregation can lead to a better performance in the downlink, reducing the quality of BCCH carriers and increasing the carrier to interference ration of TCH carriers, thanks to an increased efficiency of network functionalities such as power control. The system performance analysis carried out in section 4 considering a realistic scenario under heavy traffic load has shown that the advantages of the frequency planning with dedicated bands disappear. This is confirmed by the analysis carried out with reference to the development of frequency plans with and without band segregation considering a realistic cell layout. Based on the analysis carried out in this paper, we conclude that, under the hypothesis and assumptions considered herein, the frequency planning with band segregation can be a viable solution only in case of a small number of transceivers (e.g, high reuse factor) per cell and under low traffic load. References [1] M. Madfors et al., High capacity with limited spectrum in cellular systems, IEEE Communications Magazine, Aug 17 [2] F. Kronestedt, M. Frodigh, Frequency planning strategies for frequency hopping GSM, Proc. VTC 7, Phoenix, Arizone, USA, May , vol III, pp [3] F. Delli Priscoli, N.P. Magnani, V. Palestini, F. Sestini, Application of Dynamic Channel Allocation Strategies to the GSM Cellular Network, IEEE Journal on Selected Areas in Communications, vol, No 8, Oct 17 [4] E. Damosso, F. Grimaldi, M. Sant Agostino, Network Planning Tools and Activities in Italy, Proc. Mobile Radio Conference, Nov