EXTENDING THE SPECTRUM FOR KA-BAND SATELLITE SYSTEMS BY USE OF THE SHARED BANDS

Transcription

1 EXTENDING THE SPECTRUM FOR KA-BAND SATELLITE SYSTEMS BY USE OF THE SHARED BANDS Barry Evans, University of Surrey, Guilford, GU2 7XH, UK, Paul Thompson, University of Surrey, Guilford, GU2 7XH, UK Abstract The spectrum available in the exclusive Ka-bands has been shown to be insufficient to meet the demands of broadband access by satellite post This paper presents the work of the FP7 project CoRaSat in validating the use of the extra spectrum in the shared Ka bands for use by satellite. It is demonstrated that in the 17.3 to 17.7GHz band shared with Broadcasting Satellite Service (BSS) feeder links that throughout Europe only areas of less than 2% of the coverage would need additional interference mitigation. In the 17.7 to 19.7GHz band it has been shown that the use of a data base system of Fixed Service (FS) interferers connected to a resource allocation system at the satellite gateway allows use of virtually all of the additional spectrum and provides a capacity increase of four times that by use of the exclusive band alone. In the 27.5 to 29.5GHz uplink band the use of the High Density Fixed Satellite Service (HDFSS) portions should provide sufficient bandwidth without any mitigation but that if needed a data base plus resource allocation scheme provides similar capacity improvements to that in the downlink. 1. Introduction The demand for higher rate and reliable broadband communications is accelerating all over the world. Within Europe the Digital Agenda sets a target for universal broadband coverage of at least 30 Mbps across the whole of Europe by 2020 and 100 Mbps to at least 50% of the households [1]. Fixed connections and cellular cannot alone meet this target, particularly in the rural and remote areas but also in some black spots across the coverage. In these latter regions satellite broadband delivery is the only practical answer as satellite will cover the whole territory. Some recent studies of the roll out of broadband have shown that up to 50% of households in some regions will only have satellite available as a means of accessing broadband and thus 5-10 million households are potential satellite customers [2]. Current Ku-band satellites do not have the capacity to deliver such services at a cost per bit that makes a business case and thus the satellite community has turned to High Throughput Satellites (HTS) operating at Ka-band and above. Examples of early HTS Ka-band satellites dedicated to such services are the Eutelsat KaSat [3] and VIASAT 1 [4]. These satellites employ multiple (around 100) beams using four fold frequency reuse over the coverage area to achieve capacity of the order of 100 Gbps per satellite. The latter is limited by the exclusive spectrum available to the Fixed Satellite Service (FSS) of 500 MHz in both the up and downlinks and this limits the feasible user rates to Mbps. Thus looking ahead to the increased user demands we have to look to larger satellites (maybe up to a Terabit/s [5, 6]) and to more spectrum. Moving up to Q/V bands has already been suggested for feeder links but for user terminals the additional expense is not considered desirable so we return to the problem of getting more usable spectrum at Ka-band. The Ka-band exclusive bands for satellite are 19.7 to 20.2 GHz in the downlink and 29.5 to 30 GHz on the uplink. In these bands FSS terminals can operate in an uncoordinated manner, which means that they do not have to apply for and be granted a license by the national regulators, provided they meet set performance characteristics. The issue in other parts of the Ka-band is that the spectrum is allocated, not just to FSS but also to fixed links (FS) and to BSS (uplinks for broadcast satellites) as well as mobile services (MS). In these so- called shared bands the different services need to co-exist and this is usually done by the process of coordination. For example a larger gateway or feeder link may use this band but is coordinated and then licensed to operate and receive protection from interference from other service users.

2 Within Europe the CEPT [8] have adopted decisions that expand those of the ITU and produce tighter regulation as follows; GHz: the BSS feeder links are determined as the incumbent links but uncoordinated FSS links are also permitted in this band GHz: FS links are considered incumbent but FSS terminals may be deployed anywhere but without right of protection GHz: CEPT provide a segmentation of the band between FSS and FS portions as shown in Figure 1. Within each segment there is a specified incumbent but for instance FSS terminals can operate in FS portions provided they do not interfere with the incumbent FS. Figure 1 CEPT GHz Band Segmentation The interference scenarios considered in our study are shown in Figure 2. Figure 2 Interference Scenarios 2. A Data Base Approach The aim of this analysis is to determine maps of excess interference from BSS/FS incumbents at potential FSS sites across the coverage area. Then to evaluate the percentage of the coverage area per country that would be potentially impacted by any excessive interference. From this the area not affected by interference can also be determined. Here we use data bases of BSS stations for scenario A and of FS stations for scenario B. The data on the positions and the characteristics of the BSS and FS are generally held by national regulators and these need to be available for a database system to work. For scenario A the number of BSS uplinks in Europe is small and thus a database system is similar in magnitude to that of TVWS. However for scenario B the number of FS links runs into the tens of thousands and the database is much more complex. For the case of the UK, current data bases for BSS and FS were made available by OFCOM UK. For other countries we have used the data bases from the ITU BR-IFIC [7] with the exception of Poland for FS where additional data was available on the Internet. The information of a real interferer database is interfaced to an interference modelling engine which uses ITU- Recommendation P [8] procedures plus terrain and climatic databases. This is the latest version of the ITU Recommendation that contains a prediction method for the evaluation of path loss between stations. ITU-R P includes all the propagation effects on the surface of the Earth at frequencies from 0.1 GHz to 50 GHz. In addition, other factors which affect interference calculation,

3 such as terrain height, bandwidth overlapping are also considered in the proposed database approach. The typical interference threshold determined is based on the long term interference which can be expected to be present for at least 20% of the average year and it is set at 10 db below the noise floor. The interference thresholds for FSS reception and for FS reception are therefore: dbw/mhz and -146 dbw/mhz, respectively as given in [9] and [10]. Having determined the interference level at the FSS location (in scenarios A or B) it can be compared with the threshold indicated above. If the excess is significant then one of several measures can be taken to mitigate the interference. One such approach is to adopt an interference based resource allocation approach at the gateway where a new carrier can be assigned either in another part of the shared band where interference is acceptable or in the exclusive band. More details of the database approach are given in [11]. In the following sections we use the interference maps derived from the data bases and the modelling engine to assess the performance of a data base based system. In the geographical area analysis we determine the percentage areas available for FSS use which are below the interference limits. In the second evaluation we use a typical multi-beam satellite model for EU coverage and evaluate the capacity gain across beams over the UK and France. 3. Geographic Area Analysis Scenario A Using an appropriate BSS database, area analysis for scenario A in the UK was undertaken to investigate how much area would be affected by interference from the BSS feeder links. The band of interest is split into 10 x 40 MHz Sub-Bands (SB1 SB10) and the analysis was then conducted in each sub-band to determine the area of the contours at different cognitive zone thresholds. These mirror the approximate 40 MHz channel spacing adopted for most BSS satellites. Area analysis is based on the BSS database with the full ITU-R P model employing the terrain and climatic zones and the FSS terminal considered points to a satellite at 53 degrees E longitude. The results are for long term interference (normally 20%). It is assumed that short term interference can be covered bu appropriate Adaptive Modulation and Coding (ACM). One example of affected area at difference cognitive zone thresholds is shown in Figure 3, which represents SB1. Full data on the areas are given in the Table 1. It can be seen that in general across the sub-bands at a -155 dbw/mhz threshold less than for 2% of the area of the UK is affected by BSS feeder links and thus more than 98% of the area of the UK can be used by an FSS terminal without the need for any further action. Some mitigation of excess interference may be required in those areas affected by unacceptable levels of interference. The mitigation could be achieved by suitable site shielding, beam-forming or reallocation to another frequency that is clear at the specific location.this is very promising for future FSS deployment as the additional 400 MHz identified in scenario A ( GHz) represents an 80% increase over the current exclusive band allocation ( GHz). Although we have presented results herein for an FSS terminal pointed at a specific orbit location we have examined a range of orbit locations from the UK and the results are very similar. We have also evaluated the area in Luxembourg with similar results. It transpires that the UK results represent the most dense deployment of BSS uplinks and that the results for other EU countries should be similar or better than the UK GHz SB1 SB2 SB3 SB4 SB5 155 dbw/mhz 2, (1.06%) 1, (0.74%) 1, (0.74%) 1, (0.73%) 3, (1.56%) 145 dbw/mhz (0.30%) (0.24%) (0.24%) (0.24%) (0.40%) GHz SB6 SB7 SB8 SB9 SB dbw/mhz 1, (0.73%) 2, (1.05%) 2, (1.11%) 2, (1.03%) 2, (1.28%) 145 dbw/mhz (0.24%) (0.32%) (0.34%) (0.30%) (0.40%) Table 1 Area analysis (sq. km.) of the band GHz

4 (a) (b) Figure 3 Example of cognitive zones for the sub-band 1 ( GHz) Scenario B Unlike the situation in scenario A, the UK 18 GHz FS database used for these studies comprises many more carrier records (15,036 records) over the 2 GHz band from 17.7 to 19.7 GHz. For scenario B we perform spectrum analysis for a particular location in the UK instead of geographical area analysis across the whole of the UK to determine which carrier(s) can be used by an FSS at a specific location. This information could then be integrated with a mitigation approach such as resource allocation algorithm in the satellite network to assign the carriers to a non-interfered with frequency. Spectrum analysis results for the UK FS links at 18GHz at a specified location with latitude 52.5 degrees, longitude -0.1 degree is shown here as an example. The analysis results of the location with both LOS and full model (ITU-R P ) are shown as Figure 4 and Figure 5, respectively. (a) Interfering FS links (b) PSD based spectrum analysis Figure 4 LOS result of all UK FS links, interfering to FSS terminal at latitude of 52.5 degs and longitude of -0.1 degs. (a) Interfering FS links (b) PSD based spectrum analysis Figure 5 Full terrain model result of all UK FS links, interfering to FSS terminal at latitude of 52.5 degs and longitude of -0.1 degs

5 The FSS terminal evaluated, points to the same satellite as the previous examples which is located at 53 degrees E longitude. In each figure, a map of the links that exceed an interference level of -160 dbw/mhz is presented along with spectrum analysis as a plot of the interference power spectral density (PSD). Interference PSD is shown per MHz from 17.7 to 19.7 GHz. At this location, it can be seen that with the LOS model the interference can be from FS links much further from the location of interest and these links are ones pointing directly at the location. Only a few points with some offset and these are located very close so that interference is from their side lobes. From the interference PSD in Figure 4, it can be seen that more than half of spectrum resource from 17.7 GHz to 19.7 GHz is available (with interference below the threshold) at this location under LOS model. However, if the full terrain model is considered as in Figure 5, the number of interfering FS links dramatically decreases to less than ten, which means less than 0.1% of total FS links would cause problem at the location. Therefore the majority of the 2 GHz of bandwidth can be used by an uncoordinated FSS VSAT terminal site. Complete maps of the locations in the UK and France (see Figure 6 and Figure 7) have been produced and these can be used as input to a mitigation approach such as a resource allocation scheme which would then optimize the carrier allocation on the basis of the extra spectrum available. The resolution of the latter is 0.1 degrees in Latitude and Longitude. It was noted that although the number of FS links in the data base was large those that actually caused interference at a specific location and in a particular frequency band were quite small. It should also be noted that the available spectrum is not the same at each location and thus the data base analysis can be used to optimize the carrier allocation as a function of FSS location. Total UK FS interfering bandwidth above dbw/mhz, FSS pointing to 13E Figure 6 Map of total FS bandwidth at a FSS location for UK

6 Figure 7 Map of total FS bandwidth at a FSS location for France Area Analysis across EU countries By analysing the interference results for five data bases for countries across Europe it is possible to get an increased insight into the situation. The analysis was conducted over these regions with the full diffraction model and statistics were prepared from the results. To permit a fair comparison between the countries only the results for test points over land were included. A Cumulative Distribution Function (CDF) was produced as shown in Figure 8, for the total occupied bandwidth of the FS interferers at a point over the regions of interest. NOTE: Also in Figure 8 a second horizontal axis at the top indicates percentage of the total spectrum occupied by the FS. Table 2 summarises the spectrum use across countries. % of sites UK FRANCE POLAND HUNGARY SLOVENIA 10% 7% 3% 4% 2% 3% 1% 23% 13% 10% 14% 8% 0.1% 35% 28% 20% 41% 20% Table 2 CDF of total bandwidth of FS link interference per FSS site (% of GHz)

7 Figure 8 CDF of FS bandwidth by interferers over the five regions (for -154 dbw/mhz threshold) 4. Capacity analysis In order to evaluate the capacity gain by using the interference maps we have assumed a typical multi beam satellite coverage and characteristics [12] which provides 11 beams over the UK and 25 over France. By doing a link budget analysis across the coverage region and incorporating the terrestrial interference into the overall C/(N+I) we determine the capacity in case we use only the exclusive band and again when we use the exclusive plus shared bands. In the latter case we also use a carrier allocation scheme to reposition the carriers into the interference free spectrum [13]. The results are shown in Table 3. System gain scenario Downlink: coverage area in France / U.K. Uplink: coverage area Finland / Slovenia Exclusive band only Capacity = 168Gbps using the exclusive band [ GHz] Capacity per beam = 1,1Gbps Using the exclusive band [ GHz] Exclusive band + shared band with Cognitive radio techniques Capacity = 845 to 884Gbps using the exclusive band + cognitive radio techniques in shared bands [ GHz] (scenario A) and [ GHz] (scenario B) Capacity per beam = 5,6Gbps Using the exclusive band + cognitive radio techniques in the shared band [27,5 29,5GHz] Table 3 Capacity increase estimates per considered coverage area assuming a typical multi-spot beam satellite configuration Capacity increase 400% capacity increase 400% capacity increase Also included in this Table are evaluations performed in the uplink for coverage in Slovenia and Finland for which we had available FS data bases in this band [14] We conclude that a significant gain can be reached with the scenario A and B frequency bands. The availability of the additional frequency bands for scenario A [ GHz] and B [ GHz] would achieve an additional throughput of over 400%, which more than quadruples the overall system throughput. On the return link, with the introduction of the scenario C [ GHz] in addition to the exclusive return link band [ GHz], the throughput has been assessed as well. The usage of a proposed power and carrier allocation technique, which manager the return link transmissions, is such that the overall throughput at the input of the FS receivers is limited to the ITU limit defined. The overall assessment demonstrates on the basis of available FS link deployment data over Finland and Slovenia in these bands [ GHz] that a 400% increase in capacity can be expected over the scenario of using the exclusive band only. As noted earlier scenario C ( GHz) up link

8 may not be needed in the shorter term except for the HDFSS portion which is already protected for satellite use. More work in this scenario is needed over wider coverage. 5. Spectrum sensing The data base approach is a relatively static evaluation of the interference and to incorporate a more dynamic approach we need to adopt sensing at the terminals. In the CoRaSat project we have investigated classic methods of spectrum sensing applied to the Ka-band terminals and found all of them to have some implementation issues. However a system based on SINR estimation and use of a Data Aided SNORE estimation algorithm using knowledge from the pilot blocks in DVB-S2 has shown to be feasible. The sensing operation is less complex than conventional systems and via simulation has been shown to provide comparable outputs to the data base system. The scheme has also now been implemented in the laboratory and shown to be able to distinguish the noise from the interferer in the SINR measurements. Thus it can be incorporated into the terminals and the SINR reported back to the gateway for use in the carrier allocation scheme as in the data base system [15,16]. The sensing scheme can be used stand-alone where no data base information is available or in combination with the data base outputs to improve on the overall accuracy. 6. Conclusions and the Way Forward Using BSS data bases available to us we have shown that only very small areas around BSS stations would be adversely affected. In fact, in the UK and Luxembourg which represent the most dense deployments of BSS, less than 2% of the country area is affected. It is likely to be less for other countries. Thus by completing interference maps across Europe operators will be able to advise users in the shared bands if they are likely to need further mitigation techniques. The latter could be site shielding or the use of beam forming. Thus a further 400MHz of spectrum can be used over 98% of EU area. We do not foresee the use of dynamic data base approaches in this band. Using FS data bases available for several EU countries we have shown that large percentages of the 17.7 to 19.7 GHz band is available at most locations. For 99% of the sites evaluated greater than 80% of the 2GHz spectrum is available. However it is not the same white space at all locations. Thus a data base of FSS spectrum availability is needed to be available at the satellite network gateway. This can then be interfaced with a carrier allocation scheme to assign the carriers into the white spaces or if not available into the exclusive band. Using the data base of spectrum availability generated in the project we have used a multi beam satellite model for Europe and calculated the capacity by evaluating the link budgets across the beams. For beams covering the UK and France for which we had more reliable data bases we calculated that the capacity increase was four times that of the exclusive spectrum alone. In other words virtually the whole of the shared spectrum could be used to increase the capacity. Taking the area and spectrum analysis together with the capacity analysis this provides compelling evidence that such systems will provide additional capacity for the down link. In cases where data base information is not available interference at the terminals needs to be assessed at the terminal in situ. Classical methods of spectrum sensing such as energy detection and use of cyclostationary properties of the signals have been investigated and found to be rather difficult to apply in the satellite case. Within the project we have evaluated using a measure of the SINR from each terminal which is already transmitted back to the gateway as part of the DVBS2 ACM scheme. Using a SNORE algorithm it has been shown that interference can be detected in the presence of noise and thus a change of interference can be used with the carrier allocation scheme to optimise the allocations. The only requirement here is to build the software into the carrier allocation scheme at the gateway. Thus this scheme could be used stand alone or in conjunction with the data base approach to enhance its performance. For scenario B as in A where it proves impossible to reallocate carriers and total interference is always above the threshold we can employ other interference reduction mechanisms. The simplest is to use site shielding as this may only mean a small change of position of the terminal or some physical shielding material. We have also investigated a beam forming terminal that nulls the interference. The data base, spectrum sensing and resource allocation schemes have been demonstrated in a laboratory demo [17] and shown to operate satisfactorily; the interference maps have been interfaced to the carrier allocation scheme and the SNIR algorithm implemented and both work well. A Standards document, known as the SRDoC [18], based on the project is in the process of being published by ETSI. The outputs of CoRaSat have been fed into CEPT groups SE-40 and FM 44 and are part of a current consultation on regulatory implementation of the schemes [19]. In particular a report has been produced on a way forward for the data base information collection and is being progressed by CEPT.

9 All of the key components for operating in the extended spectrum bands have been put in place and demonstrated. The band from 17.3 to 17.7GHz can now be incorporated into future satellite designs and the cognitive zones made available to potential users. For the 17.7 to 19.7GHz band CEPT need to now confirm the way forward for producing the interference/spectrum availability information. Terminal and gateway manufacturers can build on the demonstration by Newtec within CoRaSat to produce equipment. Satellite operators need to then order the satellites using the extended bands. 7. Acknowledgements The authors would like to acknowledge the contributions of partners in the CoRaSat project to this paper. In particular we wish to thank the University of Bologna for work on sensing and the University of Luxembourg for the work on the resource allocation and the capacity analyses. In addition we also thank Newtec for the work on demonstrating the system in the laboratory. References [1] A Digital Agenda for Europe, FCC , European Commission COM 245, Brussels, Tech. Rep., [2] EU FP7 Project BATS, available at: [3] H. Fenech, E. Lance, and M. Kalama, KA-SAT and the way forward, Ka-Band Conference, Palermo, Italy., Tech. Rep., [4] Highest-capacity communications satellite, communications-satellite/. [5] P. Thompson, B. Evans, L. Castenet, M. Bousquet, and T. Mathiopoulos, Concepts and technologies for a terabit/s satellite, in Proceedings of SPACOMM-2011 (best paper award in 2011), April 2011, Budapest, Hungary. [6] A. Kyrgiazos, B. Evans, P. Thompson, P. T. Mathiopoulos, and S. Pa- paharalabos, A terabit/second satellite system for european broadband access: a feasibility study, International Journal of Satellite Communi- cations and Networking, vol. 32, no. 2, 2014, pp [7] ITU-R terrestrial BRIFIC, available: R/index.asp?category=terrestrial&rlink=terrestrial- %brific&lang=en. [8] Recommendation P : Prediction procedure for the evaluation of interference between stations on the surface of the earth at frequencies above about 0.1 GHz, International Telecommunication Union, Tech. Rep., [9] Methods for the determination of the coordination area around an earth station in frequency bands between 100 MHz and 105 GHz, ITU Radio Regulation Appendix 7, International Telecommunication Union, Tech. Rep., [10] Recommendation F.758-5: System parameters and considerations in the development of criteria for sharing or compatibility between digital fixed wireless systems in the fixed service and systems in other services and other sources of interference, International Telecommunication Union, Tech. Rep., [11] W. Tang, P. Thompson, and B. Evans, "A Database Approach to Extending the Usable Ka Band Spectrum for FSS Satellite Systems," The 7th International Conference on Advances in Satellite and Space Communications (SPACOMM) 2015, April 2015, Barcelona, Spain [12] CoRaSat Deliverable D 3.4 ( [13] Resource allocation for cognitive satellite communications in Ka band ( GHz S.Sharma, E. Lagunas, S.Maleki, S.Chatzinotas, J.Goetz, J.Krause and B.Ottersten; IEEE - ICC 2015 Workshop CogRaN-Sat [14] E. Lagunas, S. K. Sharma, S. Maleki, S. Chatzinotas, J. Grotz, J. Krause, and B. Ottersten, "Resource Allocation for Cognitive Satellite Uplink and Fixed-Service Terrestrial Coexistence in Ka-band," th International Conference on Cognitive Radio Oriented Wireless Networks (CROWNCOM), April 2015, Doha, Qatar

10 [15] V. Icolari, A. Guidotti, D. Tarchi, and A. Vanelli-Coralli, "An Interference Estimation Technique for Satellite Cognitive Radio Systems,"Â 2015 IEEE International Conference on Communications (ICC), pp , 8-12 June 2015, London, UK [16] A. Guidotti, V. Icolari, D. Tarchi, A. Vanelli-Coralli, S. K. Sharma, E. Lagunas, S. Maleki, S. Chatzinotas, J. Grotz, J. Krause, E. Corbel, B. Evans, and P. Thompson, "Spectrum Awareness and Exploitation for Cognitive Radio Satellite Communications," 2015 European Conference on Networks and Communications (EuCNC), pp , 29 June-2 July 2015, Paris, France [17] CoRaSat Deliverable D 4.4 ( [18] ETSI TR v , Sytsem reference document (Srdoc); Cognitive radio techniques for satellite communications operating in Ka band [19] ECC Report 232 Compatibility between FSS uncoordinated receive Earth Stations and the FS in the band GHz