Downlink Capacity of UTRAN reusing Frequencies of a DVB-T Network with Negligible Influence on DVB-T Performance

Transcription

1 ownlink apacity of TRN reusing Frequencies of a V-T Network with Negligible Influence on V-T erforance Jörg Huschke ) ; Wolfgang Rave 2) ; Thoas Köhler 2) ) Ericsson Research, Ericsson Eurolab eutschland GbH Neueyerstrasse 5; 94 Nürnberg; Gerany Joerg.Huschke@eed.ericsson.se 2) resden niversity of Technology - Mannesann Mobilfunk hair for Mobile ounications Helholtzstrasse 8; 62 resden; Gerany rave@ifn.et.tu-dresden.de bstract-the paper investigates the possibility of reusing frequencies of a terrestrial digital video broadcasting (V-T) network for a MTS Terrestrial Radio ccess Network (TRN) with negligible influence on the V-T perforance. frequency division duplex (F) TRN is deployed such that it can reuse all frequencies of the V-T network in each V-T frequency reuse cluster in the downlink, with a predefined level of (negligible) interference fro the TRN to V- T. The downlinks of both networks can thus coexist in the sae frequency spectru, thereby increasing the spectral efficiency. The uplink of TRN-F is assued to operate in traditional non-coexistence and is therefore not considered. lose to % of the capacity of a non-coexisting TRN are achievable for typical TRN cell sizes, with only arginal increase of outage in the V-T network. I. INTROTION Terrestrial igital Video roadcasting (V-T) is being gradually introduced in Europe, replacing analogue TV in the frequency band MHz until the year 2. In the sae tie frae, the deand for 3G services is expected to increase beyond what can be served in the spectru currently already designated or under discussion to be designated to 3G systes. In particular the downlink capacity of the current spectru plan will then not suffice for the huge deand for asyetric, downlink biased services that are anticipated to doinate in the future. This otivates investigations in the possibilities to introduce downlinks of 3G systes into further parts of the spectru, in coexistence with other systes. The frequency band below GHz, to a large extent used for TV broadcasting, is especially attractive for 3G services, since it perits in addition better perforance at high velocities of obile receivers and better rural coverage than higher frequency bands. This paper shows that it is possible to operate the downlink of the MTS Terrestrial Radio ccess Network in Frequency ivision uplex ode (TRN-F), specified by 3G, on the sae frequencies that are concurrently used by a V-T network. frequency reuse pattern of a V-T network is considered as given and a TRN is deployed such that it can reuse all frequencies of the V-T network in each V-T frequency reuse cluster in the downlink, with a predefined level of (negligible) interference fro the TRN to V-T. Instead of V-T, ost assuptions and conclusions of this paper also apply to other broadcast networks. Section II introduces the cellular layout and frequency pattern of the coexistence scenario. The calculation of the TRN capacity is derived fro interference distributions in section III. fter an overview on the siulation odels in section IV, results for the TRN capacity and V-T outage are presented in section V. brief extension of the scope to V-T single frequency networks follows in section VI. The issue of inter frequency handovers, necessary because coexistence requires ultiple frequencies for a single TRN coverage, is addressed in section VII, before closing conclusions are drawn in section VIII. This work was sponsored by the OMR research project [] of the Geran Ministry for Research and Education. II. OEXISTENE SENRIO V-T transitter sites are assued to be positioned on a grid of equilateral triangles, resulting in hexagonal cell shapes. Since the focus is on the operation of V-T in for of a ulti frequency network (MFN), adjacent transitters use different frequencies. The solid hexagons in Figure (left) for an exaple of a cellular layout of V-T with a frequency reuse of 7. The using a certain (arbitrarily selected) frequency group are highlighted by thick lines. Light gray (green) lines connect adjacent using this frequency group. These lines for equilateral triangles. TRN cell cluster shift vector V-T co-channel ells with the sae letter G use the sae frequency F E G G F E F E V-T TRN Figure : Superiposed layouts of V-T and TRN cell clusters; left: acroscopic view; right: zooed section round the triangle centers, a TRN cell cluster is arranged of the sae size as the V-T, indicated by

2 hatched hexagons in Figure left. Each TRN cluster consists of ultiple, all using the sae frequency group as the V-T in the corners of the triangle, as shown in Figure right. y placing the cluster centers on the centers of the equilateral triangles the axiu interference fro any TRN cell in the cluster to a V-T co-channel cell is iniized. In this exaple, the resulting TRN cluster structure can be considered as a shifted copy of the V-T cell structure. The coexistence concept exploits the fact that a digital broadcasting syste requires a frequency reuse factor larger than, typically 7 to 3. This is required even if single frequency networks (SFN) are used, which are addressed in section VI in this paper. In the described coexistence scenario, attention has to be paid to any inter syste interference. Figure 2 illustrates the received powers fro V-T and TRN along a path through co-channel. ue to the 6 ties larger cell density of TRN over V-T in this exaple, the power per TRN cell can be by 3d saller than for V-T, still leading to the sae received power at the cell edges of either network. TRN (k) V-T sae propagation loss exponent sae received power at borders of TRN and V-T V-T IR at cell border argin between V-T and MTS interference power Figure 2: Scheatic graph of received power of single (--) and ultiple ( ) TRN and V-T ( ) Interference fro V-T to TRN will in general not degrade the quality-of-service (QoS) the users will experience, because the transit power control of TRN will ensure that all user connections will achieve the targeted signal to interference ratio (SIR). The interference fro V-T will, however, reduce the capacity of each TRN cell, i.e. the nuber of concurrent connections that can be served. The capacity reduction is the saller, the saller the TRN cell is. Thereby, the concept akes use of the fact that TRN coonly will use uch saller cell sizes than V-T. In contrast, V-T has no eans of power control available, and therefore the interference fro TRN ay need to be liited to a level that is negligible to V-T. Given a V-T network, with specific frequency reuse factor and cell sizes, and given the propagation conditions for signals fro TRN, the interference fro the TRN is only depend- The carrier frequency raster of V-T and TRN is assued to be equal. ent on the axial transit power of each TRN base station, called Node-, and proportional to the Node- density. In the scenario of Figure 2, the interference fro TRN at a V-T cell edge is still below that of the closest V-T co-channel cell. Since the transit power of a Node- increases with the traffic load of that Node-, the interference fro TRN to V-T can be liited by restricting the load per TRN cell, using an appropriate access control echanis related to each cell of the obile syste. For a given Node- density that eans that the capacity per TRN cell will be soewhat reduced, in coexistence with V-T, copared to a non-coexistence scenario. s long as the reaining capacity justifies investents of additional transceiver equipent, the reduced capacity ay be acceptable. III. RELTIVE TRN ITY The investigated downlink capacity easure for TRN is derived in the following. The TRN transit power control adjusts the power for each user equipent (E) to reach a target SIR X tar : G X tar () I G : pathgain between E and serving Node- I : total effective interference at E Neglecting coon control channels, the total power of the Node- is the su of the powers for all M Es. ssuing there is no liit on the powers for individual Es, and that all users ai for the sae X tar, then it follows, using Eq.(): M X M tar I G If the nuber of Es is large, the su in Eq.(2) can be approxiated by the expectation value: X M E{ I/ G} for M >> (2) tar (3) If this equation is solved for M, it can be interpreted to be the capacity of a TRN cell for a given axiu power of the Node-. Since M is inversely proportional to X tar, and since we are interested in a capacity easure that is only dependent on the characteristics of the radio conditions I and G, Eq.(3) is solved for the noralized capacity easure M X tar : 2 ( ) M X tar E{ I / G} (4) can be expressed as in [2] in ters of the TRN inter cell interference I (and finally of the TRN inter cell to intra cell interference ratio F ) and the orthogonality factor α, 2 Soft handover is not considered in the capacity calculation, since [2] indicates that the effect on the relative capacity (defined below) is negligible.

3 but instead of the theral noise, the V-T interference I is considered, because it is doinant. I is expressed by the equal power of all V-T transitters and the cuulative pathgain G fro those to the Es: E I G E α + + G G α + F G α + I G + G E G + I Since the TRN are all of equal size, the distribution of G is the sae in each TRN cell. In contrast, G G,b depends on the position of the cell b relative to the V-T transitters. F depends on the nuber K of per cluster, and on the considered cell b, i.e. F F,b (K ), since the TRN inter cell interference increases with the nuber and relative position of TRN neighboring a considered cell b. If,b is the sae for all TRN, then b is different for each TRN cell b. Since it is generally desired that the cell capacity should be the sae in each cell ( b const.), for the considered hoogeneous networks, it is necessary that,b is adapted individually for each cell. enoting the cuulative power of all in a cluster by, this power can be approxiated for large K by: (5) K { } (6) E,b b sing this equation, then Eq. (5) can be rewritten as: I,,,, b G b α + E + (7) b, / K G / K G For convenience of coparison of with the noralized capacity of a standard TRN that is not interfered by V-T, the ratio R / is used in the presentation of results. can be calculated fro Eq. (7) for. In the vicinity of V-T transitters, G can becoe very large. In order to avoid that the sall fraction of such locations excessively reduces, the largest % of the distribution of the ter in the expectation in Eq. (7) is excluded, which eans that those % of the Es requiring the largest power are denied service (blocked). IV. SIMLTION MOELS The distributions of G, G and of the V-T IR is deterined by siulations considering a grid of user locations. ntennas at the transitters and at the receivers are onidirectional. ropagation is odeled at 8MHz. For V-T the propagation curve of Figure in [6] for transit and receive antenna heights of 5 and, respectively, was converted for a portable outdoor receiver at a height of.5. The curve is defined for distances larger than k and resebles a power law with exponent 5.5 in the range up to 2k. elow k the curve is extended by a power law with an exponent γ 3.3. G is the su of the pathgain to cochannel and to adjacent channel transitters, whereby for the latter the adjacent channel suppression (S) given in Table is taken into account. Therein the saller S is the worst case, belonging to the unfiltered odulation spectru, whereas the larger S assues soe reasonable filtering. ll S values assue ideal rectangular receiver filters. For TRN the pathgain over a path distance d is odeled for a base station antenna height of 5 above roof top as 3 : G / d lg( d / k) (8) Within TRN and V-T, terinals are served by the base station transitters experiencing the largest pathgain. Table : Siulation paraeters araeter V-T TRN adjacent channel suppression to V-T: 3d [4] / 45d TRN: 32d [4] / 47d V-T: 47d [5] TRN: 49d [5] bandwidth 8MHz 5MHz theral noise power - d -98 d cell radius 2 k paraeter transit power [EIR] 4 kw paraeter orthogonality α -.4 log-noral fading standard deviation 5.66 d uncorrelated between paths (equivalent to 8d for correlation of.5) V. RESLTS For a frequency reuse factor N3 and equal power for each of the K 6 TRN, Figure 3 shows that the noralized capacities b of the are siilar, except for one cell, that suffers fro strong adjacent channel interference (I). The cell capacities can therefore be equalized by noralized capacity per cell b.5.5 5% blocking S: 45d 2% blocking % blocking / -d TRN cell nuber b in cluster Figure 3: TRN cell individual capacities for equal powers 3 This odel overestiates the pathgain for d>25k, because then the pathgain in d actually increases nearly linearly with d. This range is, however, not relevant in the considered siulation scenario with 3 interfering cell tiers.

4 a oderate adaptation of the cell individual powers,b, except for the cell with strong I. For this cell, in practice the range ay have to be reduced in addition, but for the siulation it has been assued that this is not necessary. For a frequency reuse factor N3, Figure 4 shows the relative TRN capacity R /. With increasing K, R rises already at a saller power ratio / /, because for the saller TRN, less power is required for the sae per-cell capacity. R / reuse factor N 3 pattern type hex S 45 d / TRN/V [d] 7 Figure 4: Relative TRN capacity versus power ratio; for varying nuber K of TRN per V-T cell. For a given /, R increases with the TRN cell density. This is a favorable dependency, because areas with high cell densities are generally those where large capacity is required, and that is where large extra capacity achievable by coexistence with a broadcasting syste would be ost welcoe. In Figure 5, R is plotted versus / for a nuber of cobinations of the reuse factors N, the S and the frequency pattern types [3]. The types denoted hex and hex2 have hexagonal co-channel cell syetry, but are different in respect to the adjacent channel cell arrangeent. Hex axiizes the adjacent channel cell distance within each network, which can lead to overlap of TRN and V-T on adjacent channels in coexistence. This overlap is avoided by hex2 patterns and thereby the sae R is reached for saller / than for hex. In the pattern denoted ono (for onoclinic syetry) the co-channel distances between TRN and V-T are increased, resulting in larger R than for hex [3]. Figure 6 shows that the V-T outages corresponding to the configurations of Figure 5 increases with /. The outage is defined as the fraction of locations with IR saller than a target IR, whereby the latter is chosen such that an outage of 5% results for / and hex. This is a generally envisaged outage percentage for the considered portable outdoor reception of V-T. The table in Figure 6 reveals that for saller N, the 5%-IR of V-T for / decreases. orrespondingly the V-T transit power can be reduced, too, while still aintaining the desired 5%-NR to 5%-IR ratio of about 2d. Therefore, less absolute power per TRN cell is required for the sae / than for large N. The V-T outage is alost independent of K, since the cuulative power eitted fro all TRN in a cluster is kept constant. Therefore only the case for K 6 is shown in Figure 5 and Figure 6, corresponding to a cell radius of 2.5k. R / ono S45d reuse 3 3 hex2 2 hex 9 hex 3hex. cluster size TRN/V [d] 7 hex 4 hex Figure 5: Relative TRN capacity versus power ratio for a selection of scenarios V Outage (5 th perc.) pattern 4-hex hex hex hex hex hex2 S:3d S:45d / 5%-IR [d] ono ono cluster size K 6 u / TRN/V [d] 3 hex2 S 45 d reuse 3 7 hex 4 hex 3 hex 9 hex 2 hex,alt Figure 6: V-T outage versus power ratio for a selection of scenarios For N3, S3d and hex2, the V-T outage is increased to % already for /, because in this pattern, adjacent use adjacent channels, which leads to increased interference within V-T. For S45d, this effect is negligible, but in this case, the R curves are siilar for the hex and the hex2 pattern, so that overall the hex pattern outperfors the hex2 pattern. In the ono-pattern, the co-channel distance in V-T is reduced along one direction, leading to an increased outage of 7% already for /. For N4, a R of only about % is achievable for arginal V-T outage increase. However, this reuse factor is not suitable for V-T anyway, because the 5% percentile IR is with only 6.7d not optial for V-T. N9 to N3 are

5 suitable candidate reuse factors for the considered V-T MFNs. lready for N9, 9% relative TRN capacity are achieved for a V-T outage increase fro 5% to only 6%. For N3 and the hex2 pattern with S45d, this relative TRN capacity is achieved already for /.. This corresponds to a power of 66W (EIR) per TRN cell, which is in the order of transit powers envisaged for noncoexisting. VI. EXTENSION TO V-T SINGLE FREQENY NETWORKS In the previous section it has been shown that the coexistence becoes difficult for decreasing reuse factor of V- T. Therefore it is of interest to consider the coexistence concept for a V-T eploying Single Frequency Networks (SFNs), i.e. a reuse of over a nuber of neighbored transitters. In contrast to TRN, adjacent V-T transitters ay only use the sae frequency if they broadcast the sae inforation. If V-T is to provide regionally varying content, the SFNs need to be of liited extent. Such an SFN can be considered as a cell, served by ultiple transitters. Even without regionally varying content, the size of a cell covered by an SFN needs to be liited, at least for the generally preferred V-T odulation types 6QM and 64QM, due to self interference between the signals fro reote transitters within the SFN. t least SFNs covering adjacent regions need to use different frequencies, leading to a iniu theoretical reuse factor of 3 between SFNs. In practice a iniu reuse factor of 4 or 5 is necessary [7], with 7 being a typical value. Fro the results found for a V-T MFN, soe conclusions can be drawn for SFNs. The locations facing largest interference fro cochannel SFNs and fro TRN are those at the fringe of a SFN. Still assuing onidirectional transitters, the effect of aggregation of signal power fro ultiple transitters of the SFN cell serving a considered V-T receiver in such locations can be neglected. oparing further a SFN with a MFN of the sae reuse factor and with the sae cell size, then the reuse distances are naturally the sae. Therefore, the doinating effect of SFNs is that the distance to the serving transitter is reduced by a ratio K, where K is the nuber of transitters per SFN. ssuing now that the SIR X at the cell border is given by γ / 2 X ( / R), (9) where is the V-T cochannel reuse distances, R«is the V-T cell radius and γ is the pathloss exponent, then the reuse factor N S required for SFNs to achieve the sae X as for MFN, can be calculated fro the reuse factor N M for MFNs: N S N M / K, () This equation relates the MFN results found for a reuse factor N M to those that can be expected for a SFN with reuse factor N S, provided that both reuse factors are large. Otherwise, if N S is only in the order of 3 or 4, then the utual interference will be underestiated, because then soe TRN will be adjacent to cochannel V-T. Further studies are necessary to show if coexistence of TRN and V-T SFNs with practical reuse factors of about N7 are feasible. VII. INTELLIGENT TRN INTER FREQENY HNOVER etween TRN cell clusters, an inter frequency handover is necessary in the basic concept described in section II. Such handovers are undesired between, since they iply overhead that is on the expense of data throughput. Network initiated inter frequency handovers between transceivers within the sae cell, however, cause saller overhead. Inter cell inter frequency handovers can be avoided if at the borders between clusters have at least one frequency in coon. Even if only the frequencies shared with V-T are available, this can be achieved by adding one frequency of the adjacent clusters to these. sing appropriate techniques, the TRN Radio Network ontroller can coand all Es in the area between adjacent belonging to adjacent clusters to perfor an inter frequency handover to the coon frequency of those adjacent. On this frequency, Es benefit fro soft handover even across cell clusters. VIII. ONLSION oexistence of the TRN downlink with a digital video broadcasting network below GHz has shown to be feasible in soe considered scenarios of practical relevance for the exaple of V-T. More than 9% of the capacity of a noncoexisting TRN are achievable for TRN cell sizes below 2.5k, with only arginal increase of outage in the V-T network fro 5% to 6%. The nuber of available carriers to a TRN cell in coexistence with V-T according to the presented concept is equal to the nuber of carriers per V-T cell. In the frequency range MHz that is in any countries intended for the introduction of V-T, about 4 to 6 carriers per V-T cell are envisaged to be practical. This approach provides the additional TRN capacity that will be required for converged services with digital broadcasting. IX. REFERENES [] OMR project page: [2] Kio Hiltunen, Riccardo e ernardi: WM apacity Estiation; VT2 spring; pp ; Tokyo; May 2, Japan [3] W. Rave et. al.: Geoetric haracteristics of and SIR in Two oexisting ellular Radio Networks using the sae Frequencies; European Wireless 22; Florence; February 22 [4] ETSI EN V.2. (999-) igital Video roadcasting (V); Fraing structure, channel coding and odulation for digital terrestrial television; Fig. 2 [5] 3G TS 25.4 v34 (2-9); TR (S) F; Radio transission and Reception, Table 6.3 [6] IT-R: VHF N HF ROGTION RVES FOR THE FREQENY RNGE FROM 3MHz TO MHz ; Rec. IT-R.37-7;995 [7] R. rugger: Kanalbedarf und Spektruseffizienz flächendeckender V-T-Sendernetze; Technischer ericht Nr. 62/98; Institut für Rundfunktechnik; 998 (in Geran)