Inter national J our nal of Electrical and Computer Engineering (IJECE) V ol. 4, No. 3, June 2014, pp. 314 321 ISSN: 2088-8708 314 A Unified Call-to-Pray er Framew ork in Muslim W orld Naeem Al-Oudat and Abdel Ilah Alshbatat Dept. of Electrical and Computer Engineering, T afila T echnical Uni v ersity , Jordan Article Inf o Article history: Recei v ed Feb 8, 2014 Re vised Apr 5, 2014 Accepted Apr 22, 2014 K eyw ord: Unified call-to-prayer Distrib uted mosques Sound pressure le v el Logical Fermat point Local amplifier -g ain setting algorithm ABSTRA CT In man y Muslim countries there are man y sounds that are fired at nearly the same time via loudspeak ers. This sound is a call-to-pr ayer (Azan). Azan is fired from the so-called mosques in man y countries where, these s mosques are still using its o wn tim ing to trigger such call and its o wn amplifier g ain re g ardless of other mosques in the re gion. This results in an out of sync call-to-prayer firing and a v ery noisy and distracting mix of sounds in man y places at the same re gion. In this paper , a unified call-to-prayer frame w ork is proposed that sheds light on these issues and gi v es solution directions for the abo v e mentioned issues in the currently used systems. Copyright c 2014 Institute of Advanced Engineering and Science . All rights r eserved. Corresponding A uthor: Naeem Al-Oudat Dep. of Electrical and Computer Engineering, T afila T echnical Uni v ersity P .O.Box 179, T afila, 66110, Jordan. +962777526844 naeemodat@ttu.edu.jo 1. INTR ODUCTION In man y Muslim countries, there is a v ery well-kno wn sound that is fired v e times a day . This sound is the call-to-prayer or Azan, which is a timely fired call for the v e prayers namely F ajir , Duher , Asr , Magrib and Isha [1]. This sound is fired from each Muslim mosque (Masjid). What determines ho w much sound one can hear (Azan soundscape) in an y place is ho w f ar one is from Masjids. Masjids positions are determined mainly based on the con v enience of access for the habitants in an y re gion. A soundscape is defined as the o v erall sonic en vironment in a re gi on [2]. In an y soundscape in Muslim countries, Azan is fired from each mosque using loudspeak ers that are directed in man y directions on top of the minarets of the mosques. The number of mosques in a square kilometer is roughly not less than four . Each one of the mosques fire the same sound at the same time, for that reason an y clock that is out of sync in an y of the mosques causes a noisy and not so pleasant sound at prayer times. Further , each mosque usually sets its amplifiers to the highest g ain independently of other mosques which results in a v ery loud Azan in some places. These tw o issues (out of sync. clocks and highest amplifiers g ains) can be solv ed or at least mitig ated if a proper setting of a unified call-to-prayers is applied, where g ains of mosque’ s sound-amplifiers are set appropriately and clocks in the mosques are synchronized. In this paper , a unified call-to-prayers frame w ork is proposed that pro vides a solution to the abo v e tw o issues. The set of mosques in an y re gion is a dist rib uted real-time system that cooperati v ely achie v es a common goal; namely announcing the prayer times to the residents in that re gion. Ho we v er a unified call-to-prayer i s not a ne w idea, it has been applied in man y cities in Muslim w orld; Amman in Jordan is using an FM transmitter [3], Emirates is using satellite broadcasting [4, 3], and Cairo in Egypt where the y are using a common transmitter [5, 3]. The unified-call-to-prayer presented in this paper is a no v el solution that considers setting amplifiers g ains not only firing a unified Azan. The rest of this paper is or g anized as follo ws. A Unified Call-to-Prayer Frame w ork is presented and a problem is formulated in Section 2.. In Section 3. neighbor disco v ery paradigm is discussed as an important step in proposed frame w ork. A heuristi c algorithm to set proper g ains for mosques in an y re gion is presented and discussed in Section 4.. Section 5. pro vides a proof of concept via e xperimental study . Finally , Section 6. concludes the paper and gi v e a future research directions. Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 315 2. UNIFIED CALL-T O-PRA YER FRAMEW ORK In this section, the frame w ork of unified call-to-prayer is gi v en and a proposed solution is discussed. a b c c A m 1 m 2 m 3 m 4 c 1 c 2 c 3 c 4 (a) S ystem Model (b) Modified Syst em Figure 1. A system of 4 mosques with their (a) random co v ering areas and (b) a better settings of their co v ering areas 2.1. Ov er view of the Framew ork As sho wn in Fig. 1(a), the set of mosques is a group of distrib uted sites each has its o wn range of sound broadcasting to its neighbors. The dotted circles represent this range which is the border of desired Sound Pressure Le v els. There are se v eral o v erlapped re gions where the sound pressure le v el is the sum of more than one source. These re gions are sho wn in the figure, A, B and C. In these re gions the sound is pretty much a mix of dif ferent sound sources which gi v e unpleasant feeling at these re gions which is proportional to the number of sources heard. Fig. 1(a) sho ws only four mosques which is only a simplified sample of the actual distrib ution of mosques. These unpleasant feelings in these re gions result from an y or all of the follo wings: (1) time dif ferences in Azan firing, (2) se v eral persons who fire the Azan. These tw o issues can be solv ed by calling to prayers at the same time and by the same person. Further the range circles can be recalculated to reduce the o v erlapping areas. More acceptable ranges are depicted in Fig. 1(b). 2.2. Framew ork Details and Pr oblem Statement T o achi e v e the required goal of reducing the area of the o v erlapped areas, this problem can be look ed at as an optimization problem. The goal is to minimize the area of o v erlapped re gions while meeting the conditions of satisfying the residents in these re gions. One of the measures used to decide the satisf actions of the residents in these re gions is the sound pressure le v el (SPL). SPL at an y distance is proportional to the in v erse of the dist ance from the source. Hence if SPL p 1 at distance r 1 is kno wn then SPL p 2 at r 2 [6] is gi v en for open en vironment in Eq. 1. p 2 = p 1 20 l og 10 r 2 r 1 (1) Se v eral remarks can be dra wn when e xamining the problem at hand: R1. There are n nodes in a desired area A , where each node (minaret of the mosque) is equipped with a transcei v er , and an omnidirectional loudspeak er (se v eral similar loudspeak ers that co v er 360 around the minaret). R2. The link between an y tw o nodes can be af fected with se v eral parameters; T emperature, W ind V elocity , dif fraction properties and Humidity . The borders of the SPL co v erage areas of the tw o nodes shrink or e xpand based on these parameters. Sound propag ation is af fected mainly by the ground surf ace temp, where direction of the A Unified Call-to-Pr ayer F r ame work in Muslim W orld (Naeem Al-Oudat) Evaluation Warning : The document was created with Spire.PDF for Python.
316 ISSN: 2088-8708 sound propag ation bends more to w ard the colder en vironment. As of wind v elocit y , the sound tends to bend to w ard the lo west speed block of air when the sound propag ates in the wind direction and bends to w ard the highest speed block of air when the sound p r op a g at ion is not in the same direction of the wind. Sound t ends to reflect from hard obstacles when obstacle size is lar ge compared to sound w a v elength and tends to bend around when obstacle size is small as compared to sound w a v elength. As of humidity parameter , dry air attenuates sound more than humid air does. Further the required SPL at an y place is less when the background noise in the re gion is lo w . R3. When looking at the area as a whole there will be re gions that are co v ered by more than one source. R4. The lack of time synchronization between mosques results in time lapses between Azans heard in o v erlapped re gions. Azan is composed from dedicated w ords; consequently , the audiences in the o v erlapped re gions hear dif ferent w ords from dif ferent sources. R1 imposes the constraint of co v ering the desired area A . R3 gi v es the objecti v e of our solution for the problem, which is area minimization of the o v erlapped re gions. R3 and R4 impose the need of a unified call to prayer and time synchronization between mosques. While R2 sho ws the problem dynamics, where the radii of the co v erage circles are influenced by the mentioned parameters e xcept wind v elocity , where there is no sense in considering the shape of the co v erage area as a circle an y more. F or the case in Fig. 1(a), the problem can be e xpressed as a linear program (LP): Let OA denotes the sum of the o v erlapped areas. Then OA = a + b + c , And hence LP can be written as: Algorithm 1 LP formulation of UCtP problem minimize OA such that: OA m 1 m 2 m 3 m 4 + c 1 + c 2 + c 3 + c 4 = A In light of the abo v e LP , the problem at hand requires an enumeration of all possible o v erlapping between the co v erage circles then the optimal solution is the one that gi v es a smaller o v erlapped area. This means that the comple xity gro ws e xponentially as the number of nodes increases. Since an optimal solution needs a lot of space and time to minimize the o v erlapped areas, which are indeed not feasible due to the constraints and dynamics of this problem, a f aster solution is needed. Therefore a heuristic algorithm is propos ed, in this paper , to pro vide a f ast suboptimal solution to the problem at hand. The algorithm is based on a local decision that is tak en in each node based on the state of the neighboring nodes. The algorithm starts by e xchanging state describing messages between the nodes, and then each node will decide what g ain le v el will be used for its amplifier . Locally Gain setting Based on Neighboring States Algorithm (LGBNSA) is the name of our algorithm. In the follo wing section we pro vide the details of this algorithm. 3. NEIGHBOR DISCO VER Y F or an y node to disco v er its neighbors, disco v ery beacon with a highest g ain will be broadcasted periodically v e times a day ahead of Azan times by delta t. T o satisfy the pre vious goal, we assum ed that each node has its o wn GPS recei v er for time synchronization between nodes, and its priority to start beacon broadcasting is based on its ID, as sho wn in Fig. 2. Each beacon carries the ID of the broadcasting node. F ollo wing this procedure each node is capable of recognizing all its neighbors. F ailure to recei v e beacons from other nodes is considered as an indication that the node has no neighbors. Once this happen, this node should switch to its highest g ain during scheduled Azan firing. Neighbor disco v ery algorithm can be classified to tw o types based on ho w frequent it is called; short and long term disco v ery . Long term neighbor disco v ery can be periodically called for a period of 24 hours, while short term disco v ery is called v e times a day . An y node that is heard by node i when it transmits its beacon is added to the neighbor table. T o calculate the distance of that node from node i a sequence of transmissions is initiated starting by the highest po wer then starting decreasing that po wer until it is not heard. At that instant the transmission g ain is translated to a logical distance between the tw o nodes. This distance is used in calculating the logical coordinated of the mo ving node. F or each of the border nodes (that does not ha v e an y node in its co v erage circle from the border direction) tw o dummy nodes (radius of the co v erage circle of a dummy node is zero) are added to neighbors table. IJECE V ol. 4, No. 3, June 2014: 314 321 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 317 F 3 F 2 F 1 I D 2 0 I D 1 0 I D 1 Figure 2. T iming diagram for setup and Azan calling periods 4. LOCALL Y GAIN SETTING B ASED ON NEIGHBORING ST A TES ALGORITHM (LGBNSA) T o handle the problem using the proposed algorithm, tw o cases arise: 1. Number of neighbors for each node is fix ed. Based on the result obtained from long term disco v ery type. 2. Number of neighbors is not fix ed which is v arying according to short term disco v ery type. Each node has what is called a vicinity matrix that includes all nodes of distance 2R or less a w ay . Based on triangu- lation which means all possible triangle s are formed from the node and its neighbors. After forming these triangles a Fermat point is calculated that gi v es radii of the co v erage circles for n odes on v erte x es of the tria n gl e. These radii guarantee the co v erage of the triangle formed from these nodes. K eep doing this will e v entually co v er the whole re gion while k eeping the o v erlapped areas to the minimum possible. Note that R is not fix ed and it does not represent a ph ysical dist ance on earth, thus the other tw o v erte x es other than the one at focus are dynamically mo ving a w ay or close to rel ati v e to the one at focus. This phenomenon is a logical consequence as a result of the abo v e mentioned parameters. Based on this the node at focus sends beacons to ot her nodes in its adjacenc y matrix to determine ho w f ar a w ay the y are. Then a ne w logical coordinates for these nodes are formulated and implemented in calculation of Fermat point. Fig. 3 clarifies this idea. Figure 3. Logical Coordinates of nodes j and k as seen from node i The ne w point of the virtually mo ving node can be simply calculated from the line equation, assuming the mo v ement is only happen along the ph ysical line between the nodes. While it is possible for the node to sho w an illusion of c o or d i nates that are not on the same line, we did not consider it in this paper and we reserv e that for a future w ork. Let r denotes the distance between the tw o nodes n i at point ( x i ; y i ) and n k at point ( x k ; y k ) . Let d denotes the logical distance between the ne w location of node k at point ( x k ; y k ) and node i . Then ( x k ; y k ) can be calculated as in Eq. 2. y k = y i + d y k y i r ; x k = y k y i + y k y i x k x i x 1 x k x i y k y i (2) Similarly , Eq. 2 is used t o find the logi cal coordinates of node j , ( x j ; y j ) . T o calculate the radius for each of the v erte x es ( ( x 1 ; y 1 ) , ( x 2 ; y 2 ) and ( x 3 ; y 3 ) ) under consideration, we need to find the Fermat point which is done according to the w ork in [7]. Thei r equations (3) for finding Fermat point ( x F ; y F ) is repeated here for completeness of the w ork. x F = K 1 K 2 K 3 2 S p 3 d x 1 K 1 + x 2 K 2 + x 3 K 3 ; y F = K 1 K 2 K 3 2 S p 3 d y 1 K 1 + y 2 K 2 + y 3 K 3 (3) where, S = j x 1 y 2 + x 2 y 3 + x 3 y 1 x 1 y 3 x 3 y 2 x 2 y 1 j A Unified Call-to-Pr ayer F r ame work in Muslim W orld (Naeem Al-Oudat) Evaluation Warning : The document was created with Spire.PDF for Python.
318 ISSN: 2088-8708 r j l = q ( x j x l ) 2 + ( y j y l ) 2 for f j ; l g 2 f 1 ; 2 ; 3 g K 1 = p 3 2 ( r 2 12 + r 2 13 r 2 23 ) + S K 2 = p 3 2 ( r 2 23 + r 2 12 r 2 13 ) + S K 3 = p 3 2 ( r 2 13 + r 2 23 r 2 12 ) + S d = 1 p 3 ( K 1 + K 2 + K 3 ) Then for node i the distance from Fermat point, r iF , is calculated as: r iF = p ( x i x F ) 2 + ( y i y F ) 2 (4) Assuming that r 1 = 1 , p 2 = p F , and using the result of abo v e equation (4), p i = p 1 can be found from Eq. 1. Algorithm 2 Locally Gain setting Based on Neighboring States Algorithm (LGBNSA) Input: node i and its neighbors Output: node i with adjusted radius R i   max f 0, max. distance from dummy nodes if an y g f or all nodes in neighbor table of i do Construct a triangle Compute the distance from i to Fermat point of the triangle if R i is less than the computed distance then R i   computed distance end if end f or In this paper , where the goal is to shed light on the e xisting problem, we made some assumptions: Neighbors are fix ed and kno wn to an y of the nodes. The re gion that contains the mosques is a flat area, i.e., there is no mountains or v alle ys in the re gion. Mosques sound le v els are the same on all directions and form a circle centered at the specified mosque. 5. EXPERIMENT AL STUDIES In this section, we present mosques transmission radii for actual mosques on earth. Then we apply the proposed algorithm on that distrib ution of the mosques and compare the results. T o our kno wledge, no researchers ha v e been discussed this problem before. Fig. 4(a) sho ws the distrib ution of mosques in Ainalbaida re gion that lies in the southern part of Jordan. The residential area of this village is 1800 m eters wide by 3100 meters long. Eighteen mosques are distrib uted almost randomly in this area, i.e., not according to the sound le v els distrib ution around the mosque rather than the con v enience of access for the residents. T able 1 reports the con v entional coordinates of the v erte x es of the mosques. Note that the distances between mosques in this coordinate system is a representati v e system of the actual coordinate system. The mosques are numbered in order as Fig. 4(b) sho ws; the origin of coordinates is placed in the bottom-left corner of the figure. Please, bear in mind that the positions cannot be sw apped to an y ne w location and the coordinates are estimated with respect to the (x,y) reference frame. Fig. 4(b) is generated by performing a 2D Delaunay triangulation [9] on a set of eighteen mosques. The resulting diagram consists of twenty nine triangles distrib uted as sho wn in Fig. 4(b). Delaunay triangulation has the property that the circumcircle of an y triangle in the triangulation contains no mosque in its interior . The centers of the circumcircles are then connected to each other in which the y all produce a V oronoi diagram [10] which is a dual of the Delaunay triangulation. T o apply LGBNSA, the neighbors of the mosques are as sho wn in T able 2. Note that, in this table an y neighbor that is f ar a w ay by more than 2R (R considered here as 200) is dropped from the list of neighbors and hence an y triangle formed with this neighbor is also dropped from triangles used in calculating Fermat points. F or e xample IJECE V ol. 4, No. 3, June 2014: 314 321 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 319 (a) Di strib ution of mosques in Ainalbaida re gion [8] 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 (b) x,y Coordinates and Delaunay triangulation of Ainalbeada mosques [8] Figure 4. Mosques in Ainalbaida re gion (a) Distrib ution and (b) triangulation the triangle formed by mosques 1, 3, and 6 in Fig. 4(b) is not considered in an y further calculation because distances between 1 and 6; and 3 and 6 are greater than 200. Note that for an y mosque where its neighbors list is empty , its radius is the highest (200 in our e xample). Fig. 5(a) sho ws the circles that are dra wn around each mosque and represe n t its co v ering re gion. Fig. 5(b) sho ws the current co v ering circles of the mosques, where each mosque uses its max am p l ifier g ain (equi v alent to 200 in our case). W e sho wed only some of them in order to k eep the figure as clear as possible although each of the mosques has the same circle centered on it. Using the information in T able 2, the sum of radii squares is around 372333. While for the radii in Fig. 5(b) the sum of their squares is 720000. The number for the calculate d radii is small compared to the current system. This number reflects the sum of sound le v els in the re gion co v ered by the gi v en mosques, i.e., proportional to the areas of the co v ering circles. That means the o v erlapped areas are lar ger than in current system from that calculated by the proposed algorithm. This number is also proportional to the po wer consumption by the mosques amplifiers. 6. CONCLUSION AND FUTURE W ORK Call-to-prayer in Muslim w orld is an important timely fired sound. This sound is triggered v e times a day in a noisy and distracting w ay if not correctly synchronized and unified. In this paper a no v el frame w ork of unified call-to-prayer w as proposed. The problem w as formulated, analyzed, and a general solution w as proposed that is dependent on setting a suitable g ain for the dif ferent mosques amplifiers. This w ork can be implemented to serv e man y other sound generating f acilities lik e public school, churches, disaster announcement. As a future w ork, the transmitting of unified call-to-prayer will be studied to achie v e the requirements of con v enience and co v ering habitants’ areas. A suitable terrain and local en vironment models can be applied to get more accurate A Unified Call-to-Pr ayer F r ame work in Muslim W orld (Naeem Al-Oudat) Evaluation Warning : The document was created with Spire.PDF for Python.
320 ISSN: 2088-8708 T able 1. XY coordinates of Ainalbeada mosques Mosque X Y 1 55 71 2 59 81 3 132 103 4 135 159 5 277 180 6 409 102 7 385 250 8 314 344 9 207 347 10 375 414 11 107 505 12 161 497 13 231 517 14 324 528 15 210 586 16 337 660 17 161 695 18 92 713 T able 2. Neighbors and calculated radii of Ainalbeada mosques M i Set of neighbors ( j ; k ) Radius 1 (2,3) 11.5 2 (1,3), (3,4) 72.7 3 (1,2), (2,4), (4,5) 77 4 (2,3), (3,5), (5,9) 95 5 (3,4), (4,9), (9,8), (8,7), (7,6) 135.5 6 (5,7) 100 7 (5,6), (5,8), (8,10) 107 8 (5,7), (7,10), (10,13), (13,9), (9,5) 91.5 9 (4,5), (5,8), (8,13), (13,12), (12,11) 155 10 (7,8), (8,13), (13,14) 117.5 11 (9,12), (12,15), (15,17), (17,18) 179.3 12 (9,11), (11,15), (15,13), (13,9) 59 13 (9,8), (8,10), (10,14), (14,15), (15,12), (12,9) 152 14 (10,13), (13,15), (15,16) 84.5 15 (11,12), (12,13), (13,14), (14,16), (16,17), (17,11) 98 16 (14,15), (15,17) 118 17 (16,15), (15,11), (11,18) 108 18 (11,17) 49 distances between mosques in the neighborhood. Further , an y loudspeak er on the minaret of an y mosque can be tak en into account which produces a non-circular co v ering re gion. REFERENCES [1] T . S. Lee, “T echnology and the production of islamic space: the call to prayer in sing apore, Ethnomusicolo gy , v ol. 43, no. 1, pp. 86–100, 1999. IJECE V ol. 4, No. 3, June 2014: 314 321 Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE ISSN: 2088-8708 321 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 (a) Calculated radii of the mosques based on the pro- posed algorithm [8] 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 (b) Co v erage circles of the mosques in Ainal- baida [8] Figure 5. Mosques and their co v ering re gins. (a) Calculated using the proposedlgorithm and (b) actual co v ering circles [2] J. D. Porteous and J. F . Mastin, “Soundscape. J ournal of Ar c hitectur al and Planning Resear c h , 1985. [3] “http://www .uaeinteract.com/docs/ministry announces unified call system/13899.htm, last accessed: Decem- ber , 7, 2013. [4] “http://www .a wqaf.ae/service.aspx?lang=en&sectionid=19&refid=1348, last accessed: December , 7, 2013. [5] “http://www .e gyptindependent.com/ne ws/one-v oice-man y-mosques-call-prayer -unified-muezzins-defy- go v ernment-orders, last accessed: December , 7, 2013. [6] D. Bies and C. Hansen, Engineering noise contr ol: theory and pr actice . T aylor & Francis, 2009. [7] A. Y . Uteshe v , Analytical solution for the generalized fermat-torricelli problem, arXiv pr eprint arXiv:1208.3324 , 2012. [8] “T afila, jo. (03 feb . 2014). google maps. google. retrie v ed from https://maps.google.com/maps/ms?msid= 203417662707219167682.0004e6f a615a01ba6136a&msa =0&ll=30.784244,35.598621&spn=0.049699,0.090895. [9] D.-T . Lee and B. J. Schachter , “T w o algorithms for constructing a delaunay triangulation, International J ournal of Computer & Information Sciences , v ol. 9, no. 3, pp. 219–242, 1980. [10] F . Aurenhammer , “V oronoi diagramsa surv e y of a fundamental geometric data structure, A CM Computing Sur - ve ys (CSUR) , v ol. 23, no. 3, pp. 345–405, 1991. A Unified Call-to-Pr ayer F r ame work in Muslim W orld (Naeem Al-Oudat) Evaluation Warning : The document was created with Spire.PDF for Python.