Reusability of Primary Spectrum in Buildings for Cognitive Radio Systems

Transcription

1 Reusability of Primary Spectrum in Buildings for Cognitive Radio Systems Meng-Jung Ho, Stevan M. Berber, and Kevin W. Sowerby Department of Electrical and Computer Engineering The University of Auckland, New Zealand Abstract Areas inside buildings may have the potential to provide secondary radio operations in primary radio spectrum because building structures can shield signals and reduce interference. Reuse of spectrum in sufficiently shielded buildings may assist in satisfying the great demand for wireless services in urban areas. This paper considers the feasibility of reusing primary radio spectrum inside buildings. While previous research has presented analyses of the interference caused by outdoor secondary users (SUs), in this paper we determine the reduction of interference provided by building structures from signals emitted by indoor SUs. It is recognised that: not all buildings have the same construction; not all buildings have sufficient radio shielding capability; and not all parts of a building might be suitable for SU operation. The COST-231 building penetration loss model is used initially to simulate radio signal leakage from SUs within a building. The impact of building shielding on primary system interference is analysed and verified with Monte Carlo simulation. The benefits of restricting SUs operation zone within a building are also presented. If SUs are only permitted to operate in indoor zones which are well shielded from the outside primary users, then there is considerable potential for spectrum sharing, even close to the primary system. I. INTRODUCTION Most of the radio frequency spectrum below 30 GHz has been allocated for a wide variety of radio services. While some spectrum is allocated for public use such as Wi-Fi, Bluetooth etc., other spectrum is allocated exclusively for licensed operations. The public spectrum is becoming heavily used due to high and increasing demands for public wireless services. The current allocated public spectrum may become insufficient to accommodate the demand, and additional spectrum will be required. In contrast, the licensed spectrum is much less utilised. Spectrum utilisation measurements have been performed at various locations by several research groups around the world. The results show low rates of spectrum occupancy when both time and location are considered. Spectrum utilisation between 4.58% to 17.4% have been reported [1] [4]. The unoccupied spectrum in space and time is referred to as white space or spectrum holes [5] and provide an opportunity for the provision of additional future public wireless services. Allowing public use of licensed spectrum will result in interference between licensed and unlicensed users. The licensed users are generally referred to as primary users (PUs) and the unlicensed users are referred to as secondary users (SUs). The interference from the SUs to the PUs has been investigated widely [6] [9]. Spectrum sensing and geo-location databases are the most common techniques used to avoid creating interference harmful to the primary system [] [12]. Spectrum sensing is a technique that SUs use to detect the presence of a primary signal prior to transmission. Transmission is only permitted if the primary signal is absent. As an alternative to sensing, geo-location databases can be used to indicate when SUs may transmit safely. These databases contain information about primary systems, including transmission frequencies, power, PUs locations etc. and require that each SU is aware of its geographic location. Regardless of how the presence of a primary system is determined, it is imperative that SUs do not transmit from any location that will cause interference to the PUs. In general an outdoor SU at a particular location will cause more interference to a primary system than is caused by an indoor user at the same location. This is because building walls attenuate signals propagating through them. This attenuation is known as building penetration loss [13]. If SUs are located inside a building and the building structure provides a sufficient level of signal attenuation, the spectrum inside the building can be (re)used by the SUs without interfering with outdoor PUs. In this paper, we investigate the re-usability of primary spectrum in buildings. Figure 1 illustrates the concept of the reuse of spectrum indoors. As an example, we use the TV broadcast system as the primary system with the SUs operating inside the building. The interference signal from an SU will be attenuated by the building structure before it reaches the primary receiver, that is, the building provides radio shielding. The level of the radio shielding can vary significantly among different types of buildings. Buildings with predominantly glass panels or large windows may provide much less radio shielding than concrete walled buildings.

2 Primary signal Building TV broadcast transmitter Interference Signal Primary Receiver Fig. 1. Reuse of spectrum indoors. Direction of Primary Users Zone 1 Zone 2 Zone 3 Internal Wall and floor must be prohibited from transmitting in these areas. In this paper we investigate the aggregate interference power emanating from indoor SUs. In Section II, research work related to reusing spectrum indoors is discussed. This is followed by a survey of building penetration loss measurements in Section III. The system model used to study the feasibility of reusing the primary radio spectrum inside buildings is presented in Section IV and the results of an initial analysis using the model are discussed in Section V. Section VI concludes the paper. Fig. 2. Internal view of the Building in Figure 1. Moreover, the level of radio shielding can also vary greatly in different parts of a building. This can be seen in Figure 2 which shows the internal structure of the building in Figure 1. The building is divided into three zones. The TV receiver might experience the highest level of interference from the SUs operating in zone 1 because of their elevated position within the building. On the other hand, the interference from the SUs in zone 3 could be low due to the fact that a) zone 3 is the furthest zone away from the TV receiver; b) zone 3 is low in the building; and, c) the SU signal must propagate through more than one wall to reach the primary system. SUs in zone 2 are more likely to cause interference than the users in zone 3 but less likely than the users in zone 1. Several aspects of this concept of indoor spectrum reuse need to be considered. Firstly, the radio shielding ability of a building is a key determinant of the maximum number of SUs that can be accommodated safely within the building. It also determines the required separation distance between the PUs and the building in order for the interference to be acceptably small. Secondly, the aggregate interference from multiple buildings has to be analysed. Thirdly, operational management of SUs in various parts of a building has to be investigated. SUs operating in different parts of a building cause different level of interference to the outdoor PUs. Some areas of a building might not allow reuse of the spectrum and SUs II. RELATED WORK Most research work on interference to PUs only considers outdoor interference sources [6] [9]. Interference from indoor sources is seldom considered. In [14], Novillo et al. investigated the area inside a building within which the primary spectrum is reusable. A grid of identical multi-floor buildings was assumed in the scenario considered. Each of the buildings had four floors and each of the floors was divided into 25 identically sized rooms. The primary system operated only on the rooftops of the buildings and the SUs (re)used the same primary channel inside the buildings. The two way interference links between the PUs and the SUs were both considered in the analysis, which determined the reusable area of the spectrum inside the buildings. The COST-231 building penetration loss model was used in the analysis to calculate the interference between the users. The results showed that floor 1 (the bottom floor) provided the largest reusable area, with the smallest area being provided on floor 4 (the top floor). In [15], Obregon et al. investigated the number of channels in TV-bands available for secondary operation indoors. Only one building, one PU and one SU were considered in the analysis scenario. A TV set was assumed to be the PU and the SU was a white space device located inside the same building. Co-channel operation was forbidden for the SU and the interference to the PU was caused by the adjacent channel operations of the SU. The Keenan Motley (KM) indoor propagation model [16] was used to take account of the propagation

3 loss through the walls and the floors in the building. A rooftop antenna, cable reception and set-top antenna were considered as three separate cases in the analysis. It was found that a higher number of channels was available for the SU in the rooftop antenna and cable reception cases. In [17], Kim et al. evaluated the network capacity of both the primary system and the secondary system. The secondary network consisted of indoor and outdoor SUs. A -db indoor shadowing factor was applied to all indoor SUs to take account of the building penetration loss. The results showed that the secondary network had a significantly higher capacity if the SUs were located indoors and had knowledge of the radio environment. The contribution of this paper is that we explore the reusability of primary spectrum in buildings by investigating the co-channel interference to outdoor PUs from indoor SUs. We use the TV spectrum sharing scheme proposed by the Federal Communications Commission (FCC) as an example to demonstrate the benefit of limiting SUs operations to indoor environments []. The COST-231 building penetration model is also used for the indoor-to-outdoor interference analysis. Then, the aggregate interference at any location inside the primary system coverage area has been analysed. Analytic results are verified using Monte Carlo simulation. III. RADIO BUILDING PENETRATION Research on radio propagation into and within buildings has been widely reported. The motivation for most of the research work on this topic was to develop a reliable statistical model to predict the signal strength within buildings for cellular mobile systems [13], [16], [18]. However, very little research has investigated the signal strength outside buildings when the transmitters are indoors, i.e. radio leakage from the buildings. Penetration models can be applied to determine the radio leakage of buildings since the radio propagation channel is reciprocal. The COST-231 building penetration model is one of the most widely referenced models and is chosen for our initial analysis of indoor spectrum reuse. Figure 3 illustrates the basic geometry of the model and its parameters [19]. S is the physical distance from the outdoor antenna O to the externally illuminated wall and D is the perpendicular distance from O to the wall. d is the perpendicular distance from the wall to the indoor antenna I. All the distance parameters are in metres. The propagation loss between the antennas is given by L db = log(f) + 20 log(s + d) + W e + W Ge (1 D S )2 + max(γ 1, Γ 2 ), (1) Outdoor Antenna O S D Fig. 3. φ d Indoor Antenna I Path Loss External Wall COST-231 building penetration model Power (db) Fig. 4. Radio leakage of a 15 15m building with a transmitter located at the centre. f = 1.8, W e = 7, W Ge = 20, d = 7.5 and α i = 0.6. The received power is relative to the transmission power. Assuming 0dB reference transmission power. where Γ 1 = W i p (2) Γ 2 = α i (d 2)(1 D S )2. (3) f is the frequency in GHz. W e is the external wall loss (in db) at perpendicular penetration angle ϕ = 90, see Figure 3. W Ge (in db) is the additional loss in the external wall for ϕ = 0. W i is the loss in the internal walls in db and p is the number of penetrated internal walls. If there is no internal wall, p = 0, an additional indoor loss Γ 2 is determined with α i in db/m. Based on this model, 2-dimensional (2D) radio leakage simulations have been performed and the results are shown in Figures 4 and 5. A 15 15m idealised building is assumed for the simulation and only one indoor transmitter is considered. The figures show the power leakage pattern in a m area. Figure 4 shows that the leakage patterns are identical on every side of the building when the transmitter is at the centre of the building. Contrarily, in Figure 5, the leakage

4 Primary System Coverage Area ε R SU Operable Area Separation Area R 0 PT R x θ r SU PR Power (db) Fig. 6. System geometry of Case A model. Fig. 5. Radio leakage of a 15 15m building with a transmitter (0dB transmit power) located at the one side of the building.f = 1.8, W e = 7, W Ge = 20 and α i = 0.6. d = 1.5 from the left wall. power on the left side of the building is about 12dB higher than it is on the right side. This is a result of asymmetric placement of the transmitter in the building. The transmitter is placed 1.5m away from the left wall. Simulation results show that the location of the transmitter inside the building has a significant influence on the leakage power. Moreover, the strongest leakage direction is always perpendicular to the building walls. Hence, only the leakage in directions perpendicular to the walls need to be calculated if we only consider the strongest leakage power, i.e. the worst case scenario. IV. SYSTEM MODEL The TV spectrum sharing scheme proposed by the FCC is used to assess the benefit of reusing spectrum indoors. In the scheme, the primary transmitter (PT) is a TV broadcast transmitter and the primary receivers (PRs) are TV receivers. The geometry is shown in Figure 6. In the figure, the PT is assumed to be located at the centre of the primary system coverage area with the radius R 0. Outside this area, the primary signal is assumed to be too weak for the PRs to receive. SUs are not allowed to operate in or near the coverage area in order to avoid interference to the PRs. Hence, a separation area, the shaded region in between the coverage area and the SU operable area in Figure 6, is specified to separate the SUs from PUs. The distance between the coverage area and the SU operable area is defined as ε in this paper. The FCC specifies that this distance should be determined by the signal strength of the PT. The distance should be large enough so that the signal strength at the edge of the primary system coverage area is 30dB higher than the signal strength at the end of the separation distance. As a result, ε would be almost the same as R 0 and the separation area would be almost 2.5 times the coverage area [20]. This is a large area where the primary channel is not used by either the primary and secondary systems. Therefore, we assess the benefit of reusing spectrum indoors by allowing indoor SUs to operate in the separation area, and calculating the expected total interference power from all the SUs. A. Case A In the first case, we assume that all the SUs are randomly and uniformly distributed in a circular area with radius R and density λ, except in the primary system coverage area where no SU is allowed. Any SU outside this circle is assumed to be too far away to contribute any significant interference. All the SUs in the separation area operate only indoors. The expected total interference power at any location inside the primary system coverage area, from all the SUs can be expressed as E[I total ] = E[I 0 ] + E[I sz ], (4) where E[I 0 ] and E[I sz ] are the expected total interference power from the SUs in the SU operable area and the separation area respectively. Assuming free space propagation, E[I 0 ] is given by [9]: E[I 0 ] = R R 0+ε 2π 0 ( = πλp I ln λrp I dθ dr (Rx 2 + r 2 + 2R x r cos θ) α/2 (5) R 2 R x 2 (R 0 + ε) 2 R x 2 ), (6) α=2 where α is the propagation exponent and r is the distance between a SU and the PT as shown in Figure 6. P I is a reference power measured at a unit distance (0 metres) away from a SU and assuming all the SUs have the same P I. The expected total interference power from the indoor SUs in the separation area is obtained by applying the COST-231 penetration loss model to equation 5. All buildings are assumed to be the idealised square building

5 as used in Section III. Moreover, we also assume that only one SU is inside each building and located at the centre. If the worst cast scenario is considered, all buildings are assumed to be orientated such that they each have a wall perpendicular to the PR, i.e. ϕ = 90 in Figure 3, so that the PR will receive the strongest leakage interference. Hence, the propagation loss predicted from the COST-231 model can be reduced to a free space propagation loss plus an additional wall loss W e. In this case, 7dB of the wall loss is used as recommended in the COST-231 model for a building with typical concrete wall with normal sized windows [19]. Therefore, following the same derivation in [9], E[I sz ] across the entire primary system coverage area can be expressed as E[I sz ] = B. Case B R0+ε 2π R 0 = πλp I We 0 λrp I / We dθdr (Rx 2 + r 2 + 2R x r cos θ) α/2 (7) ( (R0 + ε) 2 2 ) R x ln R0 2 R x 2 α=2. (8) where Second Region First Region R ε/2 ε/2 PT R 0 R x θ r PR SU Fig. 7. System geometry of Case B model. l 2 = R 0 (11) u 1 = R 0 + ε (12) u 2 = l 1 = R 0 + ε 2. (13) In the second case, we consider a more complicated approach. We assume that a higher wall penetration loss for buildings at locations closer to the primary system coverage area is necessary to allow secondary operations in the spectrum without creating harmful interference. Therefore, as shown in Figure 7, the separation area is divided into two regions. The wall loss requirement in the first region, W e1, is set to db since it is further away from the the primary system coverage area, while the wall loss in the second region, W e2, is set to 15dB since it is closer. Moreover, it is considered that the higher the wall loss that is required, the fewer the number of buildings that can meet the requirement. For this reason, we reduce the SU density in the first region, λ 1, to be 50% of the density in the SU operable area. Similarly the density, λ 2, in the second region is set to 20% of the density in the SU operable area. Therefore, E[I sz ] in this case can be expressed as E[I sz ] = u2 2π l 2 0 u1 2π λ 2 rp I / W e2 dθ dr (Rx 2 + r 2 + 2R x r cos θ) α/2 λ 1 rp I / W e1 + l 1 0 (Rx 2 + r 2 + 2R x r cos θ) dθ dr α/2 (9) ( ( 2 2 ) λ2 u2 R x = πp I ln W e2 l R x + λ ( )) u1 R x ln W e1 l 2, () 2 1 R x α=2 V. RESULTS In this section we present the feasibility of reusing spectrum indoors in the FCC s TV spectrum sharing scheme. Figure 8 shows the expected total interference power E[I total ] in four different cases: the best case, the worst case, Case A and Case B. The plots present E[I total ] in dbm against R x from 0 to 50 (i.e. 0km to 5km). This represents the interference power level as it varies across the primary system coverage area. The first curve in the figure gives the result for the interference power when only E[I 0 ] is considered. This serves as the best case result for comparison. The second curve shows a scenario when the SUs are allowed to operate in the separation area but outdoors. Not surprisingly, the interference power is the highest of all the cases considered. This is most evident at the edge of the primary system coverage area, R x = 50. In Case A, the SUs in the SU operable area are inside buildings and the total interference power is substantially lower than in the worst case result for outdoor SUs. The difference in attenuation is attributable to the additional 7dB wall attenuation for indoor SUs. However, there is a significant increase in the interference power at the edge of the coverage area in Case A. In Case B the interference power is much less than in Case A and approaches the best case interference level (in which no SUs operate in the separation region). In Case B, the indoor SUs operating in the separation area do not contribute any significant increase to the total

6 E[Itotal ](dbm) E[I SZ ]=0 W e =0dB Case A Case B R x ( 0m) Fig. 8. Expected total interference power (in dbm) in 4 different cases. P I = 0.001, λ = 0.01, R 0 = 50, R = 500 and ε = 50. W e = 7 for Case A. W e1 =, W e2 = 15, λ 1 = and λ 2 = for Case B. interference power in the primary system coverage area. This suggests that the primary spectrum in the separation area, from which FCC rules currently exclude SUs, has the potential to be used for secondary operations indoors. VI. CONCLUSIONS AND FUTURE WORK In this paper, we have presented a primary spectrum sharing paradigm which allows SUs to operate inside buildings in areas which are currently excluded from secondary operations. We used the COST-231 building penetration model to assess the benefit of reusing spectrum indoors and discovered that there is an opportunity to share the primary spectrum in a large geographic area which is currently being excluded by the FCC. By restricting SUs to operate only indoors, the primary spectrum can be reused in the separation area while still keeping the interference to the primary system acceptably low. Several other aspects of this paradigm still remain to be investigated. Different buildings may have very different shielding properties. Methods for determining the radio shielding ability of buildings need to be developed. The shielding ability of a building is a critical factor in determining the maximum number of SUs allowed to be in the building and their maximum transmission power. Moreover, an accurate localisation technique for the SUs inside a building has to be developed. This is because not all zones within a building are suitable for frequency reuse and must be excluded from SU operation. Furthermore, not only can the primary spectrum be used indoors but also the adjacent channels. Since not all buildings have the same shielding ability, the primary spectrum can be used in the building with higher shielding ability and the adjacent channels can be used in buildings with lower shielding. Lastly, the shielding ability of a building may also be deliberately increased by applying additional radio shielding materials to the building walls, such as frequency selective surfaces [21], [22]. Future buildings might be designed to provide high radio shielding in key frequency bands. In summary, there appears to be considerable potential for greater secondary reuse of primary spectrum by allowing controlled reuse with buildings in areas currently prohibited from secondary use. Accurately locating SUs within these buildings will be very important if intolerable interference is to be avoided. Greater spectrum reuse is possible in buildings that provide greater shielding, thus raising the issue of whether such buildings might be designed for this purpose. REFERENCES [1] D. A. Roberson, C. S. Hood, J. L. LoCicero, and J. T. Mac- Donald, Spectral occupancy and interference studies in support of cognitive radio technology deployment, in IEEE Workshop SDR 06., 2006, pp [2] M. A. McHenry, P. A. Tenhula, D. McCloskey, D. A. Roberson, and C. S. Hood, Chicago spectrum occupancy measurements & analysis and a long-term studies proposal, in Proc. of the 1st Int. workshop on Technol. and policy for accessing spectr. Boston, Massachusetts: ACM, 2006, [3] M. H. Islam, C. L. Koh, S. W. Oh, X. Qing, Y. Y. Lai, C. Wang, Y.-C. Liang, B. E. Toh, F. Chin, G. L. Tan, and W. Toh, Spectrum survey in Singapore: Occupancy measurements and analyses, in 3rd Int. Conf. on CrownCom 08., 2008, pp [4] R. I. C. Chiang, G. B. Rowe, and K. W. Sowerby, A quantitative analysis of spectral occupancy measurements for cognitive radio, in IEEE 65th Veh. Technol. Conf. (VTC2007-Spring), 2007, pp [5] FCC, Report of the spectrum efficiency working group, Tech. Rep., Nov [6] R. Menon, R. M. Buehrer, and J. H. Reed, Outage probability based comparison of underlay and overlay spectrum sharing techniques, in IEEE Symp. on New Frontiers in Dyn. Spectr. (DySPAN), 2005, pp [7] X. Hong, C.-X. Wang, and J. Thompson, Interference modeling of cognitive radio networks, in IEEE 67th Veh. Technol. Conf. (VTC2008-Spring), 2008, pp [8] R. Dahama, K. W. Sowerby, and G. B. Rowe, Outage probability estimation for licensed systems in the presence of cognitive radio interference, in IEEE 69th Veh. Technol. Conf. (VTC2009- Spring), 2009, pp [9] M.-J. Ho, S. Berber, and K. Sowerby, Primary system interference analysis for cognitive radio transmission power control, in 16th APCC, Nov. 20, pp [] FCC, Et docket no , second report and order and memorandum opinion and order, Tech. Rep., [11] H. Celebi and H. Arslan, Utilization of location information in cognitive wireless networks, IEEE Wireless Commun., vol. 14, no. 4, pp. 6 13, [12] S. Haykin, D. J. Thomson, and J. H. Reed, Spectrum sensing for cognitive radio, Proc. IEEE, vol. 97, no. 5, pp , [13] P. Barry and A. Williamson, Modelling of uhf radiowave signals within externally illuminated multi-storey buildings, IRE J., vol. 57, no. 6, pp. S231 S240, [14] F. Novillo, H. Galeana, R. Ferrus, and R. Agusti, Spectrum availability in indoor locations for opportunistic spectrum access in dense urban scenarios, in IEEE 69th Veh. Technol. Conf. (VTC2009-Spring), 2009, pp. 1 5.

7 [15] E. G. Obregon and J. Zander, Short range white space utilization in broadcast systems for indoor environments, in IEEE Symp. on New Frontiers in Dyn. Spectr. (DySPAN), 20, pp [16] A. Motley and J. Keenan, Personal communication radio coverage in buildings at 900 mhz and 1700 mhz, IET Electron. lett., vol. 24, no. 12, pp , Jun [17] Y. Kim and G. de Veciana, Joint network capacity region for cognitive networks heterogeneous environments and rfenvironment awareness, IEEE J. Sel. Areas Commun., vol. 29, no. 2, pp , [18] J.-E. Berg, Building penetration loss along urban street microcells, in 7th IEEE Int. Symp. on Pers., Indoor and Mobile Radio Commun. (PIMRC 96), vol. 3, Oct. 1996, pp vol.3. [19] E. Damosso and L. Correia, Digital Mobile Radio Toward Future Generation Systems. COST 231 Final Report, [20] D. Gurney, G. Buchwald, L. Ecklund, S. L. Kuffner, and J. Grosspietsch, Geo-location database techniques for incumbent protection in the tv white space, in IEEE Symp. on New Frontiers in Dyn. Spectr. (DySPAN), 2008, pp [21] G.-H. Sung, K. Sowerby, M. Neve, and A. Williamson, A frequency-selective wall for interference reduction in wireless indoor environments, IEEE Antennas Propagat. Mag., vol. 48, no. 5, pp , oct [22] M. Neve, A. Austin, E. Lai, G. Rowe, A. Williamson, and K. Sowerby, Electromagnetic propagation in structures and buildings - research at the university of auckland, in IET Seminar on Electromagn. Propag. in Structures and Buildings, 2008, pp. 1 6.