See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/235683136 DVB-T network planning: A case study for Greece ARTICLE in IEEE ANTENNAS AND PROPAGATION MAGAZINE · FEBRUARY 2009 Impact Factor: 1.32 · DOI: 10.1109/MAP.2009.4939022 CITATIONS READS 8 117 8 AUTHORS, INCLUDING: Dimitra Zarbouti Ioannis Foukarakis 26 PUBLICATIONS 45 CITATIONS 16 PUBLICATIONS 48 CITATIONS University of Peloponnese SEE PROFILE University of Peloponnese SEE PROFILE Dimitra I Kaklamani Iakovos S. Venieris 215 PUBLICATIONS 990 CITATIONS 279 PUBLICATIONS 1,312 CITATIONS National Technical University of Athens SEE PROFILE All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately. National Technical University of Athens SEE PROFILE Available from: Dimitra Zarbouti Retrieved on: 04 February 2016 DVB T network planning: A case study for Greece D. A. Kateros, D. A. Zarbouti, D. C. Tsilimantos, C. I. Katsigiannis, P. K. Gkonis, I. E. Foukarakis D. I. Kaklamani and I. S. Venieris National Technical University of Athens School of Electrical and Computer Engineering Heroon Polytechniou 9, 15773, Athens, Greece E mails: dimitris, chkatsig@icbnet.ntua.gr, dzarb, dtsiliman, pgkonis, ifouk@esd.ntua.gr dkaklam@mail.ntua.gr, venieris@cs.ntua.gr The terrestrial digital video broadcasting (DVB T) network planning is the main interest of this article along with a case study for Greece. The basic principles and the guidelines of the DVB T planning process are presented in conjunction with their application to the establishment of the Greek DVB T allotment plan. The procedures described in this article were followed by the authors during the International Telecommunication Union (ITU) DVB T planning project that concluded with the ITU’s Regional Radiocommunication Conference 2006 (RRC 06) and the Geneva (GE 06) frequency plan for terrestrial digital radio broadcasting. allotment, compatibility analysis, DVB T, FAP, interference calculation, plan synthesis, SFN 1 INTRODUCTION Traditionally, the terrestrial broadcasting planning approach involved the definition of a number of assignments, each of which consists of the transmitter site specified in terms of longitude and latitude, as well as the specific transmitter’s characteristics and antenna’s configuration. These parameters are chosen to ensure acceptable reception or ‘coverage’ of the desired service in an area associated with, and usually surrounding, the transmitter location. However, the desired coverage of the assignment was not explicitly taken into account during the development of the plan and, in principle, could not be determined until the plan was finalized. This is the approach used for the establishment of the Stockholm (ST 61) broadcasting plan. On the other hand, a plan must not only provide means to specify ones right to use a spectrum resource, but more importantly, means to protect its effective utilization. On this line of thought, it is more essential for a plan to achieve protection of known service areas, than to specify the characteristics of a number of transmitting sites. The developments in standardization of digital radio broadcasting systems have introduced new possibilities regarding the methodology of planning terrestrial broadcasting networks. The introduction of single frequency networks (SFNs), allows the synchronization of the transmitters of a confined geographic area, so that they may operate at the same frequency channel without any destructive interference occurring at the receiver. Therefore, the planning of digital radio broadcasting networks has gained a degree of flexibility, which allows transmitter characteristics and placement calibration, as more than one transmitter may be used to achieve coverage over a given area without depleting spectrum resources. This approach that has already been employed in recent broadcasting plans, such as the Maastricht 2002 terrestrial digital audio broadcasting (T DAB) plan and the Geneva 2006 DVB T and T DAB plan [1], allows the simplification of the planning process using a higher level of abstraction, namely allotments. The actual implementation of the allotment into a set of assignments can be postponed to a latter stage. In addition, this planning approach is appropriate for very large scale planning that may comprise regional broadcasting networks with different degree of design and implementation progress. The first part of this paper is dedicated to the overview of the basic principles and guidelines that can be taken into account during the planning process of a DVB T allotment plan. Due to the different conditions in terms of propagation characteristics, terrain morphology, demographics, cultural diversity and economics that can be found in various regions, it is evident that there is no standard methodology for the design and the implementation of such a plan. In the second part of the paper, we present our planning experience for the Greek DVB T allotment plan and outline the methodology and decisions taken towards its establishment. The presented allotment plan is part of the GE 06 digital terrestrial broadcasting plan [1], whose production was coordinated by ITU and consists of the broadcasting plans 118 involved countries. Throughout this paper we examine the processes of designing the allotments, conducting interference analysis and performing the distribution of available frequency channels between them. 2 DVB T ALLOTMENT PLANNING 2.1 Allotment Area Definition An allotment is a service area that is intended to be protected against interference. In order to define an allotment, three parameters must be determined; an ordered set of geographical positions called ‘test points’, that represent the allotment boundaries, a maximum value for the acceptable interference level at the allotment test points and means to calculate the outgoing interference from the allotment. This section outlines the procedure of allotment boundary definition. The primary goal of allotment area definition is to ensure its coverage with a single frequency, as SFNs exhibit self interference limitations. The 8K carrier operational mode of the DVB T system which is suitable for SFN networks, defines 896 Es useful symbol duration ( ) [2]. In order to compensate for the multi path effect present in the terrestrial broadcasting environment which leads to inter symbol interference in the presence of echoes, a guard interval (GI) is assumed. Two transmitters operating in the same SFN network transmit simultaneously the same symbol. Assuming corresponds to distance difference = ⋅ the speed of the propagating signal this (Table I). The application of a propagation prediction technique can then identify areas within the intended coverage area that may suffer from self interference. The design of allotments is also bound to non technical parameters and criteria, such as demographics and boundaries, both regional and national. The implementation of an allotment as an SFN implies the broadcasting of a common set of programs within the allotment. Consequently, allotments must be efficiently mapped to distinct demographic regions, in order to account for linguistic diversity or to reflect administrative divisions either international or within a country. Lastly, allotment areas, as service areas, should be particularly focused on areas of significant population density. The aforementioned constraints have already been considered to some extent for the implementation of the analogue TV networks. Therefore, the planning of digital radio broadcasting services with allotments can benefit greatly building on this experience. Another aspect of the allotment definition has to do with the efficient frequency reuse. There is a straightforward relation between allotment size and the relevant frequency reuse distance. As the allotment area size decreases with relation to the reuse distance the allotment network becomes more “spectrum hungry” and the amount of frequency channels that can be assigned to each allotment area while satisfying interference constraints is reduced. The spectrum resources are limited, and therefore extremely valuable, so the allotment design should receive feedback from the evaluation of the interference relations between allotments. In the framework of the RRC 06, the only suggested method for allotment definition was the Channel Potential Method [3]. This method describes a procedure for the conversion of existing analogue assignments to allotment areas for digital broadcasting. The method aims to design allotment areas that can be mapped to and exhibit a certain degree of compatibility with the assignments of the agreed analogue broadcasting plans. Thus, its application facilitates the transition from analogue to digital broadcasting, which is a fairly complex problem, as it involves not only different technical criteria, but also different transition timetables between various regions. Although the method has a significant regulatory usefulness, providing means to convert an existing and agreed analogue broadcasting plan to a digital one, its application is mostly based on geometrical criteria, so many technical concerns are raised. Overall, the method is more suitable for latticed analogue broadcasting networks of landlocked regions and cannot be seen as a global solution for all cases. 2.2 Compatibility Analysis The goal of the compatibility analysis process is the assessment of the interference constraints between the defined allotments. The abstract concept of allotment has to be associated with specific parameters to represent the required service coverage inside the allotment as well as the outgoing interference. Two general approaches can be introduced for the compatibility analysis process, each corresponding to a different progress status concerning the network design and resulting to a different level of accuracy. The initial estimation of the environment that the DVB T network will be implemented is essential in order to define the appropriate propagation model and its parameters. In the framework of RRC 06 the propagation model employed for terrestrial broadcasting is described in the recommendation P.1546 [4]. However, it is an empirical, not location specific model based on measurements that do not consider diffraction losses nor sub path attenuation. Typical planning software tools include more sophisticated field strength prediction methods that can be used in conjunction with a terrain digital elevation model (DEM) in order to obtain more accurate results. Every network planning process is triggered by the subscriber’s and operator’s needs and wishes. Therefore, an initial set of parameters that depends on these needs must be defined. These basic parameters and constraints are described shortly in the followings. Each allotment of the planning area should be associated with a reception type and a coverage quality. In DVB T the fixed, the portable (indoor, outdoor) and the mobile are the supported reception modes while the coverage quality is denoted through the percentage of locations that the desired service is provided within the service area. The decision about the reception mode must be followed by the decision concerning the desired throughput. There exist a large number of different possible DVB T systems (variants) depending on the choice of modulation type and code rate. For instance, the offered throughput in an 8MHz channel ranges from 5 (QPSK 1/2) to 30Mbps (64QAM 7/8) according to the selected DVB T variant and GI. The selection of the DVB T variant is determined taking into account both the required reception type and the desirable throughput. Apparently, the DVB T variant determines the minimum signal to noise ratio (SNR) that is required at the receiver end, while equation (1) indicates the minimum power. = 10 log ( ⋅ ⋅ where = 1.38M10 23 J/K is the Boltzmann’s constant, )+ (1) = 2900 K is the absolute temperature, is the channel bandwidth, is the receiver noise figure and case of fixed reception mode. Nevertheless, the are the feeder losses in the provides the requirement in a noise limited planning basis which is not the case for broadcasting networks. Therefore, it is common practice to introduce a margin in the link budget calculations so as to take into account the network interference potential during the planning process. In the framework of GE 06 plan an interference margin (O) of 3dB [5] was considered. Hence, the minimum required power of (1) becomes: = 10 log ( )+ ⋅ ⋅ + (2) Finally, the aforementioned choices lead to the determination of the protection ratio (PR) value [6]. In essence, the PR value stems from the and the considered interference margin, while it provides an estimation of the interfering signals level related to the wanted ones. Obviously, the PR is strictly bound to the DVB T variant choice together with the min calculated in (2). Equation (3) provides the power threshold for acceptable reception in conformity with the DVB T planning criteria. ≥∑ + (3) =1 where is the interfering power from the interfering transmitters. Note that the th transmitter and is the total number of value of (3) along with the antenna effective aperture and gain provides the minimum required field strength, ,min( ) given in dBEV/m, that apparently depends on the frequency. ! " The above paragraph indicates that there is an important number of possible planning configurations. ITU in order to minimize them and facilitate the planning process grouped them into three RPCs [5]. Specifically, the possible combinations of reception mode, modulation type, code rate and required location coverage probability were grouped according to the equivalent minimum median field strength required on the receiving location. The minimum median field strength in frequency is given by: = where is the man made noise, desired coverage probability and + + + (4) is the location correction factor that corresponds to the are the losses that must be taken into account in relation to the reception mode. In portable outdoor reception accounts for losses caused by the building height while in portable indoor reception there are additional losses for the building penetration. The # values for representative frequencies are given in [5]. The evaluation for different frequencies is conducted using a field strength correction formula [4]. $ The interference calculations are the final step for the compatibility analysis. In particular, there are two approaches that can be followed according to the degree of insight concerning the future allotment implementation as a set of assignments. 2.2.3.1 1st Approach: Specific Transmitting Sites In case that a substantial subset of the transmitting sites implementing the allotment has been determined, accurate calculations concerning both the wanted field strength and the interference potential of each allotment can be performed. In this approach a field strength prediction method that accounts for terrain morphology is essential. During the compatibility analysis the total interfering, ∑ =1 and wanted field strength, , are calculated at every test point and once again the inequality of equation (3) applies. In order to obtain a strict estimation of the maximum allowed interference we consider % and we solve the derived equality. The maximum allowable interference at any allotment test point is then given by: = − (5) When the criterion of equation (3) falls through then the incompatibility is reported to the synthesis process. It must be highlighted that the examination of equation (3) at every test point considers all potential interference sources. 2.2.3.2 2nd Approach: Reference Networks (RNs) When the specific transmitting site locations corresponding to an allotment have not yet been determined, there is a need to model the outgoing interference by alternate means. The method that was proposed by ITU and was adopted in the framework of RRC 06, involved the definition of generic network structures with geometrical symmetry and homogeneity with regard to transmitter characteristics, called RNs. The use of RNs for the compatibility analysis is bound to the use of RPCs since the characteristics of the transmitters at each RN differentiate according to the RPC. There are four kinds of RNs, each corresponding to a different wanted reception type and service area characteristics. Their detailed description can be found in [5]. In addition, the propagation model used for this approach of interference calculation is the empirical P.1546. Fig. 1 depicts the RN placement in order to conduct interference calculations between two allotments. The interfering field strength is evaluated on each allotment test point and the largest calculated value is considered as the overall interference to the wanted allotment. The interfering field strength is derived by the summation of the individual field strengths produced by each transmitter. It is apparent that this method does not take into account multiple interference situations where the wanted allotment receives harmful interference levels due to the addition of interfering field strengths from more than one allotments. The compatibility status (compatible or incompatible) of each pair of allotments is examined by formula (6) and is reported to the synthesis algorithm. ≤ In the above inequality the #( − − − ) is evaluated with equation (4), (6) is the location correction factor of the interfering allotment and & (&'0) is the antenna discrimination [7] in the case that the wanted allotment has fixed reception (RPC 1). Note that inequality (3) does not apply for this approach since it refers to a multiple interference assessment scenario. Equation (5) is used instead in order to determine the maximum accepted interfering field strength. The utilization of RNs allows the calculation of the minimum separation distance between two co channeled allotment areas (reuse distance) [8]. The reuse distances can be calculated by employing the interference potential curves of P.1546 and by taking into account the maximum acceptable interference ( () for each of the wanted and interfering allotment combinations, i.e. solving the equation (5) for the reference frequency of the band under consideration. 2.3 Plan Synthesis Plan synthesis is the procedure of allocating frequency channels to the defined allotment areas. It can generally be seen as a variation of the frequency assignment problem (FAP). The basic FAP consists of assignment constraints, interference constraints and an objective or optimization criterion. The frequency assignment model typically involves a predefined set of frequency channels f and a set of nodes n requiring frequency channels. For every node , a subset of the available frequencies f is specified, from which a number of frequencies ( ) (the multiplicity of the node) must be assigned to it. The problem is usually represented by a graph. Each vertex represents a node and each edge represents an incompatibility between two nodes for a given pair of frequency channels. A more suitable representation for computer processing is that of a x matrix, where is number of nodes. Each element , ) of this matrix is an integer that corresponds to the required channel separation between the and ) nodes. For digital radio broadcasting systems, adjacent channel interference constraints need not be taken into account due to the low protection ratios that apply [6]. Consequently, the above representation is simplified to binary interference constraints. In fact, most models of the frequency assignment problem found in the literature use binary constraints as input [9]. However, this approach does not take into account multiple interference. In the case of DVB T planning, multiple interference must be considered since the DVB T system is characterized by rapid signal degradation from perfect to no reception within a narrow dB margin. On the contrary the analogue TV systems involve different impairment grades [6] [10]. The feasibility of possible frequency distributions can be assessed taking into account the results of the compatibility analysis. The synthesis algorithm in the case of DVB T allotment planning should account for additional features. Firstly, existing analogue assignments within the allotment area may determine specific channels to be assigned to the allotment. Moreover, the channel distribution between allotment areas must be balanced in order to avoid unfair channel distributions where very few channels are assigned to ‘difficult’ allotment areas. This constitutes a problem of many FAP algorithms, observed in [11]. Due to these requirements, algorithms employing heuristic techniques are more appropriate. The more frequently meta heuristics used are the Simulated Annealing [12], the Taboo Search [13] and the Genetic Algorithms [14]. Another heuristic approach involves greedy methods and more specifically sequential algorithms [15]. The procedure led by ITU in RRC 06 involved the submission of requirements from Administrations (input requirements). This resulted in a synthesis algorithm with the objective to maximize the number of requirements accepted for entry in the GE 06 Plan [16]. A sequential algorithm was employed consisting of two steps. In the first one the ordering of the requirements was conducted. In the second step each requirement was assigned a channel from its available channel list. A total of three approaches for the first step and of five for the second one was defined. Out of the fifteen different plans produced by the possible combinations, the one satisfying the highest number of requirements was selected. The success of this method relied heavily on the pre coordination of the input requirements by the Administrations. Specifically, ITU provided means to eliminate the detected incompatibilities between requirements taking into account the results of the pre coordination procedures (administrative declarations). 3 GREEK CASE STUDY Greece is located in the Southeastern Europe with a population of approximately 11 million. The Greek terrain morphology is extremely complex and diverse, as mountains (almost 60 percent of the land) coexist with the third largest coast line in the world and almost 200 islands, half of which are inhabited. Approximately 97% of households rely on the terrestrial analogue TV broadcasting networks, while the remaining 3% are satellite subscribers. The existing terrestrial analogue broadcasting network has been built in an ad hoc manner with little central planning, a situation which has led to the scarcity of spectrum resources as well as significant interference problems over a number of service areas. These facts, add significant difficulties in the establishment of a DVB T plan, as diverse propagation conditions and existing analogue broadcasting experience need to be carefully taken into account in order to establish a spectrum efficient plan that will serve adequately the population needs. The planning procedure was conducted on behalf of the Greek Ministry of Transport and Communications, which is the authority responsible for the country’s representation to the ITU concerning spectrum regulation and management. The ICS Telecom of ATDI was used as a network planning and propagation prediction software tool. This software allowed the performance of simulations over a digital elevation model (DEM) representation of Greece with 50m resolution. The results of these simulations were used to make propagation and interference calculations and thus calibrate transmitter characteristics, define SFN networks and conduct a compatibility analysis between the defined allotment areas. 3.1 Allotment Area Definition In general, the allotment boundaries definition is a gradual process that involves several refinement steps, each one with different criteria that must be met. The steps that were followed in the case of the Greek allotment plan are: 1) Examination of the analogue transmitter network and assessment of the existing service areas 2) Definition and extraction of the “useful sites” of the analogue TV network 3) Grouping of the “useful sites” into SFNs. 4) Examination of the coverage areas of the produced SFNs and initial construction of the allotment plan. 5) Allotment partitioning, merging and resizing according to received feedback from compatibility analysis and plan synthesis processes. At first, the examination of the operating analogue network offered a good estimation of the service areas that appear in the Greek geographical area. The most important service areas represented large cities and surroundings or well separated areas by the terrain morphology. The transmitting sites that corresponded to these service areas were examined and a list of important terrestrial broadcasting locations was determined. Throughout this paper we denote these locations as “useful sites”. The majority of the “useful sites” exhibits significant spatial and demographic coverage and they are located in high altitudes comprising a relatively clear first Fresnel zone. Although the resulted “useful sites” from the above constraints were already in use by the analogue TV network, the respective transmitter characteristics for the DVB T were determined as a result of simulations performed using ICS Telecom. Table II summarizes the basic planning parameters in case of Greece. The power thresholds of the minimum required and maximum allowed interference have been derived by (3) and (5) respectively. Note that due to the selected reception type there were no feeder losses ( = 0). According to the aforementioned network parameters adequate choices for the transmitting power and the antenna configuration were made. During this procedure additional effort was made in order to reduce the outgoing interference from the envisaged allotment. Table III provides a transmitter configuration example. The importance of SFNs was highlighted in the previous section. Therefore, the “useful sites” were grouped in SFNs. The GI duration was chosen at 224Es, which corresponds to the maximum time of arrival (ToA) difference between any received signals from the transmitters within the same SFN. This time interval corresponds to 67 km (see Table I).The initial grouping of the “useful sites” to SFNs was performed by limiting the distance between two transmitters to 67 Km, which constitutes a convenient approximation in order to take into account the aforementioned ToA constraint. It must be noted that the estimation of the distances between the “useful sites” could not allow for a smaller GI, as it would increase significantly the number of produced SFNs. The first allotment plan that was resulted by the former procedure contained 35 allotments. Nevertheless, a further examination of the SFNs was undertaken since their initial definition was based on the transmitters’ distance. The examination took place with the ICS Telecom and for a specific allotment the self inference levels did not allow the future SFN implementation. As a result the specific allotment was divided into two. We should note that the transmitter network comprised of the “useful sites” does not provide full area coverage to the majority of the defined allotments and additional planning with gap filling transmitters is required; however it is sufficient for planning purposes, for two reasons. Firstly, it provides substantial population coverage and secondly the outgoing interference from an allotment is dominantly provoked from the “useful sites”, as they have significantly larger e.r.p. and effective heights than the envisaged gap fillers. 3.2 Compatibility Analysis After the initial allotment definition the compatibility analysis took place. As it was mentioned earlier, in the framework of the RRC 06, ITU used the empirical propagation model of the recommendation P.1546 due to its simplicity and low computational effort. However, in our planning there was a need for more accurate field strength calculations, hence the Fresnel propagation prediction method was chosen. Opposite to the ITU R P.1546, which does not take into account geometrical attenuation terms such as sub path attenuation and multi edge diffraction, the Fresnel module included in the ICS Telecom provides several choices for this purpose. The diffraction losses were considered through the Deygout method [18] since it is the most widely used for this kind of planning [19]. Nevertheless, this method usually overestimates the losses in cases of many obstacles too close to each other, so the more approximate Bullington method [20] was chosen for the mountainous environment of central and northern Greece [21][22]. In order for the sub path attenuation to be considered in our studies the standard sub path attenuation method provided also by ICS was used. The sub path attenuation coefficient used by ICS is given by (7) and it is added to the total losses calculations. * =ρ⋅ (7) In (8) ρ = (#1 + # 2 + #3 + # 4 ) / # is the proportion of the total path that is located above the first Fresnel virtual ellipsoid (see Fig. 2) and is the Deygout correction term given by (8). = 20 log(75000# ) − 20 log(π +1+2 ) (8) In (8) # is the distance between the transmitter and receiver, +1 and +2 are the transmitter and receiver heights respectively and is the operating frequency in MHz. The compatibility analysis was conducted by calculating the outgoing interference from the transmitter network of each allotment to the test points of all potentially affected allotments. This resulted in two kinds of incompatibilities between two allotments, which were classified as “hard” and “soft” based on the following inequality: ≥ where + + (9) is the received field strength on each test point from the transmitters of the wanted allotment, is the interfering field strength on the same test point, protection ratio and is the relevant is a margin in dB. “Hard” incompatibilities concern allotment pairs for which inequality (9) falls through for at least one of the examined test points for = 0. “Soft” incompatibilities, on the other hand, concern allotment pairs where no “hard” incompatibility exists and inequality (9) falls through for at least one of the examined test points for = 6. The “soft” incompatibilities can additively cause the same effect as a “hard” incompatibility. Fig. 3 summarizes the ‘soft’ and ‘hard’ incompatibilities identified between the allotments of the final Greek plan for Band IV. These incompatibilities constitute the constraints to be taken into account during the plan synthesis. As stated in the previous section, the transmitter network assumed for planning did not necessarily provide sufficient coverage to the entire allotment area. Therefore, the received field strength value in some allotment test points was less than the minimum required. To compensate for this fact, equation (10) was used. = max { where , } (10) is the wanted field strength used for the compatibility analysis on each test point. 3.3 Plan Synthesis The results of the compatibility analysis are summarized in a matrix C, whose element ) is the field strength on the test point from allotment ). C contains all the necessary information for wanted and interfering field strength values on allotment test points. In order to conduct the plan synthesis an algorithm employing known meta heuristics was implemented. Details on the design and performance of the algorithm can be found in [23]. This algorithm has two stages; in the first one, it creates as many mutual compatible frequency layers as possible. A frequency layer denotes a set of evenly among all channels that are distributed allotments. In the second stage additional frequency channels are allocated without violating any of the constraints. During this stage provision is taken to preserve balance to the channel allocation between allotments. This is essential, as the heterogeneous terrain caused substantial differences between the number of incompatibilities of the allotments situated in main land and those situated in sea. An additional measure taken to facilitate the synthesis process was the unification of neighboring allotment areas containing islands. This led to a final plan of 33 allotments, illustrated in Fig. 3. Another feature of the implemented algorithm is the possibility to accept as input a predefined partial channel allocation which remains fixed during the synthesis process. This was essential in order to allow for the synthesis process to take into account the results of the coordination procedure that took place between Greece and its neighboring countries during the preparation activities for RRC 06. A simple example of the algorithm’s functionality is demonstrated visually in Fig. 4. In this example a synthesis problem consisting of 5 allotment areas and 10 available frequency channels is assumed. The graph represents the constraints. Stage 1 includes steps 1 3, while stage 2 steps 4 7. Fig. 5 displays the performance of the synthesis algorithm by presenting the CDF of the number of channels allocated to an allotment area in the case of Greece over 50 independent runs. Three scenarios are shown: Compatibility analysis performed by ITU. Compatibility analysis performed using ICS Telecom and real transmitter network. Plan synthesis ignores multiple interference. Compatibility analysis performed using ICS Telecom and real transmitter network. Plan synthesis considers multiple interference. Fig. 5 suggests that the outgoing interference prediction method based on reference networks and ITU R P.1546 propagation model is not suitable for the Greek terrain and allotment design. The majority of detected incompatibilities, which lead to the poor synthesis results, can be ignored. This can be attributed to three reasons. Firstly, the employed propagation model does not take into account terrain morphology and differentiates only between land and sea paths. Secondly, the careful selection of real transmitter sites and characteristics further reduces the outgoing interference. Lastly, RN placement as described earlier may result in placement of transmitters directly on sea (see Fig. 1), which induces additional calculated interference. Moreover, the small difference between the CDFs of the two scenarios stemmed from the compatibility analysis conducted with ICS Telecom indicates that even for the limited number of the Greek allotments the inclusion of multiple interference calculations may prevent the acceptance of invalid channel distributions. On the contrary the binary constraints scenario fails to detect these cases and produces more optimistic results. 3.4 Attiki: an implementation example The most important allotment with respect mainly to its population is the allotment Attiki which also includes the city of Athens. Fig. 6 displays the transmitter network assumed for planning purposes, as well as a possible future implementation for this allotment. As shown, the SFN network that consists of four transmitters does not provide sufficient coverage of the allotment (Fig. 6a). The SFN, comprised by 8 transmitters, clearly results to a higher percentage of allotment coverage (Fig. 6b). Note that the outgoing interference levels do not substantially increase with the addition of the four transmitters (see Fig. 6c and 6d). Specifically, no alteration to the constraints established during the compatibility analysis process takes place. This shows that the methodology followed resulted to an allotment plan that is possible to be implemented with a DVB T transmitter network that will serve the future needs of operators and subscribers. 4 CONCLUSIONS In this paper a methodology for allotment based DVB T planning that builds upon existing analogue transmitter network experience and infrastructure was introduced. Based on this methodology it is possible to define the allotment boundaries as well as assess their interference relation. The degree of flexibility induced to the allotment implementation by SFNs allows the implementation of a complete transmitter network without violating any of the initial planning assumptions concerning interference constraints. The experience of the planning in case of Greece indicates that even in difficult propagation conditions (warm sea, islands) the proper site selection and transmitters’ configuration can lead to manageable interference constraints between allotments. Additionally, topographic decoupling present in mountainous areas can be utilized for the same purpose. Lastly, the multi interference inclusion in the plan synthesis process led to a more viable channel allocation. It must be noted that the significance of the multiple interference consideration increases when the number of allotments included in the planning process increases (larger countries). 5 ACKNOWLEDGEMENTS The authors would like to thank Mr. V. Goltsios whose knowledge and experience on the analog TV broadcasting infrastructure of Greece proved very useful towards the establishment of the Greek DVB T plan. 6 REFERENCES [1] ITU R, “Final Acts of the Regional Radiocommunication Conference for planning of the digital terrestrial broadcasting service in parts of Regions 1 and 3, in the frequency bands 174 230 MHz and 470 862 MHz (RRC 06),” June 2006. [2] ETSI, “Digital video broadcasting (DVB): Frame structure, channel coding and modulation for digital terrestrial television (DVB T),” ETSI, Tech. Rep., ETS 300 744, Nov. 2004 [3] Germany (Federal Republic of), Austria, Liechtenstein (Principality of) and Switzerland (Confederation of), “A possible approach to convert analogue assignments into allotments for digital terrestrial broadcasting,” Document 56, RRC 04, May 2004 [4] ITU R P. 1546 2, “Method for point to area predictions for terrestrial services in the frequency range 30 MHz to 3 000 MHz,” August 2005 [5] ECC, “Technical Criteria for Allotment Planning of Digital Video Broadcasting Terrestrial (DVB T) and Terrestrial – Digital Audio Broadcasting (T – DAB),” ECC, Tech. Rep., Report 49, April 2004. [6] ITU R BT.1368 6, “Planning criteria for digital terrestrial television services in the VHF/UHF bands,” May 2007 [7] ITU R BT.419 3, “Directivity and polarization discrimination of antennas in the reception of television broadcasting,” June 1990 [8] ECC, “Re use distances for different combinations of reception,” WG RRC 06, Document RRC/PT2(05)061, September 2005 [9] K. I. Aardal, S. van Hoesel, A. Koster, C. Mannino and A. Sassano, “Models and Solution Techniques for the Frequency Assignment Problems,” ZIB Report, 2001 [10]ITU R BT.655 7, “Radio frequency protection ratios for AM vestigial sideband terrestrial television systems interfered with by unwanted analoguevision signals and their associated sound signals,” Feb. 2004 [11]B. Jaumard, O. Marcotte, C. Meyer and T. Vovor, “Comparison of Column Generation Models for Channel Assignment in Cellular Networks,” , &** #- + , vol. 112, pp. 217 240, Sept. 2001 [12]S. Kirkpatrick, C. D.Gelatt and M. P. Vecchi, “Optimization by Simulated Annealing,” , vol. 220, pp. 671 680, May 1983 [13]F. Glover, “Tabu Search Part I,” &. * , pp. 190 206, 1989 [14]J. H. Holland, &# * # & / , University of Michigan Press, 1975 [15]C. W. Sung, and W. S. Song, “Sequential Packing Algorithm for Channel Assignment under Cochannel and Adjacent Channel Interference Constraint,” 0 + [16]ITU R, + /, vol. 46, pp. 676 685, Aug. 1997 “The Methodology of the Planning Software: The Synthesis,” Radiocommunication Bureau, Document IPG 1/EP/2, June 2005 [17]ITU R, “The Methodology of the Planning Software: The Analysis,” Radiocommunication Bureau, Document IPG 1/EP/4, July 2005 [18]J. Deygout, “Multiple knife edge diffraction of microwaves,” & # * , vol. 14, pp. 480 489, July 1966 [19]D. I. Sirkova, and M. A. Mikhalev, “A Review of the Applicability of the Main EM Field Calculation and Assessment Models in Wireless Communications Design,” in , Nis Yugoslavia, Oct. 2002. [20]K. Bullington, “Radio propagation at frequencies above 30 MHz,” in Proc. Institute of Radio Engineers, 35, No 10, pp. 1122 1136 [21]M. Allegrettii, C. Lucianazi, R. Notarpietro and G. Perona, “Validation of Deterministic Radio Propagation Algorithms for UHF Mobile Network Planning in Mountain Environment,” in 1 +2 + & # * , Loughborough, UK, April 2007 [22]Switzerland, “Application of Different Diffraction Models in Mountainous Terrain,” Document 3K/17, Nov. 2003 [23]D. A. Kateros, P. G. Georgallis, C. I. Katsigiannis, G. N. Prezerakos and I. S. Venieris, “An Algorithm for the Frequency Assignment Problem in the case of DVB T Allotments,” Lecture Notes in Computer Science, Springer, vol. 4818, pp 549 556, under publication TABLE I MAXIMUM SIGNAL SOURCE – RECEIVER DISTANCE DIFFERENCE GI=Tg/Tu 1/4 1/8 1/16 1/32 Tg (.s) 224 112 56 28 /s (Km) 67 33.5 16.75 8.375 Figure 1: RN placement in order to conduct interference calculations between two allotment areas. TABLE II Basic Network Parameters Reception Type DVB T variant Guard Interval (GI) Implementation Margin (O) ( Protection Ratio (PR) Propagation Model Receiver height Portable Outdoor 16 QAM 2/3 ¼ 3 dB 79dBm 63dBm 16 Fresnel (Bullington) 2m TABLE III Transmitter Configuration for site Ymittos (Band IV) Tx Configuration Power (Watt) Antenna Gain Signal Modulation Antenna Diagram Analog TV 1000 12 dB PAL Analog DVB T 250 12 dB DVB 8MHz 16 QAM 2/3 Figure 2: Fresnel zones between transmitter and receiver for sub path attenuation losses calculation. Figure 3: The final Greek allotment plan along with the “hard” and “soft” constraints graph for Band IV. Figure 4: A demonstration of the functionality of the synthesis algorithm. Figure 5: The cumulative distribution function for the number of channels allocated to an allotment area over 50 independent runs. (a) The wanted field strength produced by the four “useful sites” (b) The wanted field strength produced by the four “useful sites” and four additional gap fillers (c) The interfering field strength produced by the four “useful sites” (d) The interfering field strength produced by the four “useful sites” and four additional gap fillers Figure 6: Wanted and interfering field strength for allotment Attiki with a network consisting of the four “useful” sites with which the planning was conducted and with four additional gap filling transmitters.