Implementation of Beam Formation for WMNs

Transcription

1 Implementation of Beam Formation for WMNs M. UTHANSAKUL, S. PRADITTARA AND P. UTHANSAKUL School of Telecommunication Engineering Suranaree University of Technology 111 University Avenue, Muang, Nakhonratchasima THAILAND an Abstract: - A low-cost beamformer capable of main beam an null locating is propose in this paper. This beamformer is suitable for WMNs as the positions of mesh router are fixe. The beamformer consists of 2 2 rectangular array space by λ/4 accompanie with a simple beamforming network. With multiplying constant values to output signals from beamforming network, null locating can be accomplishe. Firstly, its beamforming capability is teste through computer simulation. Then, a prototype of the beamformer is constructe an teste to confirm its performance in term of signal strength, throughput an elay. Also, the impact of position shift which mesh router is not locate on the expecte beam irection has also been investigate. Key-Wors: - Antenna array, Beamforming, Null steering, WMNs. 1 Introuction Recently, Wireless Local Area Networks (WLANs) have become an infrastructure in every builing [1]. The connection or communication between users an networks is accomplishe through Access Points (APs). In WLANs, APs communicate to each other using cables. This causes an expense an somehow introuces ifficulty in accessibility for some areas. To tackle these impairments, the iea of exploiting raio signal instea of cables has been recently propose, so calle Wireless Mesh Networks (WMNs) [2]. These networks are constitute by raio noes organize in a mesh topology. Once one noe can no longer operate, the rest can communicate to each other irectly or through one or more intermeiate noes. To this en, a raft extension of the IEEE stanar for WMNs is uner evelopment [3]. As raio signal is utilize in WMNs, co-channel interference remains a limiting factor which the system esigners have to concern. To eal with this impairment, lot of attention in the area of WMNs has been pai to smart antenna technologies. These techniques rely on beamforming algorithm to provie maximum gain at a esire irection an steer nulls or sielobes to unesire irections. The key success of smart antennas is an antenna array an a suitable signal processing unit. Fully aaptive smart antennas are able to perform the electrical beam an null steering [4]. These capabilities come with a high level of computational for signal processing unit, resulting in high expense an complexity. On the other han, switche-beam antennas being one typical type of smart antennas o not nee aitional cost an complications. In these systems [5], a number of preefine beams are forme in ifferent irections by antenna array an beamforming network. However, switche-beam antennas have the limitation of interference suppression as it cannot control nulls irections. Although this problem can be avoie when utilizing fully aaptive smart antennas, its complexity makes the concept impractical for WMNs. Therefore, this paper proposes a low profile beamformer with null locating capability for WMNs, which is practical for implementation. This beamformer provies multi-beam patterns simultaneously aroun the router of interest. In aition, interference signals can be cancelle with a straightforwar null-steering metho escribe in the paper. Moreover, the true evaluation in term of practical realization of the propose beamformer is also reporte in this paper. This paper organize as follows. After brief introuction, problem formulation in term of air interface between mesh routers is iscusse in Section 2. In Section 3, a low profile beamformer employing 2 2 rectangular antenna array an its beamforming network is etaile. To tackle the problem of interference signals from neighbouring routers, a straightforwar null steering algorithm is presente in Section 4. Section 5 shows simulation ISSN: Issue 12, Volume 9, December 2010

2 Fig.1 Configuration of WMNs employing ifferent antenna systems, (a) omni-irectional antenna (b) irectional antennas (c) propose antenna. Fig.2 Configuration of beamforming networks. results to confirm the propose concept. Moreover, the performance evaluation of switche beam antennas with null locating for WMNs is presente The prototype is constructe an teste uner IEEE a/b/g infrastructure. The performances in term of signal quality, throughput an elays are investigate. Sometimes it is impossible to locate mesh router on the beam irection of esire noe. Hence, also in the paper, the effect of position shift which mesh router is not locate on the expecte beam irection has also been examine in Section 6. Finally, Section 7 conclues the paper. 2 Beamforming Concept in WMNS Fig.1 presents WMNs in ifferent scenarios in term of air interface. Note that this paper focuses on the mesh routers, not mesh clients. This is because positions of mesh routers are relatively stable an evenly istribute within the network. So far, the concept of WMNs has been initialize with omniirectional raiation, as shown in Fig.1a. As we can Fig.3 Simulate raiation patterns of beamformer at 2.45 GHz [7]. see, interference signal from neighboring routers become interference, as strong as signal strength from the router of interest. This introuces an increase in blocking probability. Accoring to this, many works in literatures have propose to aopt the concept of using switche-beam systems an irectional antenna, respectively, in WMNs [2], [6]. As shown in Fig.1b, an antenna array is utilize at mesh routers to provie irective gain in the esire irection. The energy saving can be also achieve with this concept. However, interference signals remains in the system an their amount is relatively large for the array having high sielobe levels. This can be ecrease by utilizing a large number of antenna elements, resulting in high expense. So, controllable null locating is attractive when employing a small number of antenna elements in orer to meet the requirement of low profile systems. Fig.1c emonstrates the concept of air interface propose in this paper. As we can see, multiple beam patterns are esignable an simultaneously prouce in any given irections ISSN: Issue 12, Volume 9, December 2010

3 with a low compact size array in cooperating with a suitable beamforming network. In each beam pattern, nulls locations can be prouce in the irections of unesire signals. The etail of each part is escribe as follows. 3 Low Profile Beamformer for WMNS The beamformer consists of 2 2 rectangular array an beamforming network as shown in Fig. 2. The choice of 2 2 array configuration is ue to the requirement of a minimum number of antenna elements which is able to take responsibility for signals coming from 0 to 360 aroun the array. The array is space by λ/4 as shown in the figure. The reason is that spacing of λ/4 give lower sie-lobe level comparing with spacing of λ/2. The receive signals are elivere to beamforming network in orer to accomplish beam formation in preefine irections, simultaneously. The beamforming network is constitute by four 64 hybri couplers an a crossover. Note that the mentione beamforming network provies main beam irections at 45, 135, 225 an 315 simultaneously. Afterwars, the four outputs, from four beams, are conveye to null locator as shown in Fig.3, which is etaile in next section. 4 Null Locating Metho Nowaays, lots of effective null steering algorithms can be foun in literatures [8], [9]. Unfortunately, those methos require high level of computation which can be hanle by expensive signal processor. This is not attractive for WMNs application [10]-[11]. Therefore, this paper presents a straightforwar null locating metho which requires only multiplying some suitable coefficients at the output signals from beamforming network. The mentione coefficients can be calculate as follows. Assuming irections of incoming signals from azimuth irections, weighting coefficients at (m,n) th antenna element corresponing to the esire an unesire signals can be express as w w q ( m, n) ( m, n) = e j = e π π ( m 1) cosφ + ( n 1) π π ( m 1) cosφi, q + ( n 1) sinφi, q j 2 2 sinφ 2 2 (1) where φ an φ q are irections of arrival for esire signal an q th interfere signals, respectively. The output signals in term of esire (y ) an interference signals (y q ) from beamforming network shown in Fig.2 can be written as y y = w q = w s q total s total (2) where s total stans for signal vector which inclues esire an interference signals. Accoring to Eq.(2), the total output signal can be given by y total = y = s + q= 1 + kq y q q= 1 ( w + k w + + k w ). s q ( w + k w + K+ k w ) K (3) 1 where k q is here in efine as a interference suppressing coefficients which can be calculate as follows. With the concept of interference rejection, the 2 n term of Eq.(3) must be vanishe as the following expression. q= 1 ( w + k w + + k w ) 0 s K (4) q. 1 1 = Solving Eq.(4), the interference suppressing coefficients can be obtaine as shown in Eq.(5). 1 k1 w 1s 1 L w s1 w s1 M = M O M (5) k w 1s L w s w s The simplicity of null locating metho for this paper is inicate in Eq.(5). With simply multiplying k q at output signal from q th beam prouce from beamforming network, the q th interference signal can be simply eliminate from the systems. In practice, we can fin interference suppressing coefficient k q with information of irections of esire an interference signals, then we can follow the proceure from Eq.(1) to (5). In the circumstance of WMNs, the irections of those signals coming from mesh routers are fixe. However, the irections of signals in other wireless systems can be easily foun using some straightforwar algorithms available in literatures [12]-[14]. ISSN: Issue 12, Volume 9, December 2010

4 Fig.4 Raiation pattern of a 2 2-beamformer when the esire signal is coming from 45 an interference signals are coming from 135, 225 an Simulation Results The propose concept is teste using own evelope computer program. The raiation pattern is the key factor to inicate the beamforming capability for this paper. In the assume mesh network, there are four mesh routers locate aroun the router of interest at 45, 135, 225 an 315 from boresight irection. The router of interest has to operate to all four meshes simultaneously. Also it has to istinguish between esire signal from one mesh an interference (unesire) signals from neighboring mesh routers. The 4 monopole-antennas space by λ/4 at 2.4 GHz are employe an place as a 2 2 lattice as shown in Fig.2. Fig.4 shows raiation pattern of the beamformer forming its main beam to 45 without null locating (ash line). In this figure, soli line represents beam formation when the interference signal from neighboring mesh routers are coming from 135, 225 an 315 from boresight irection. Comparing these two results, we can see that null locator succees in interference signal suppression while preserving the main beam irection to the esire signal. The performance of the propose beamformer in some other scenarios is emonstrate in Fig.5. In this case, the beamformer nees to form its main beam to more three irections, 135, 225 an 315 (apart from 45 ). When one irection is chosen to be esire irection, the others can be viewe as interferers. As we can see in this figure, the propose systems can steer nulls to irections of interferers while pointing main beam to the esire irection. This confirms that interference signals can be eliminate from WMNs. Also, this provies energy saving which is one important issue for WMNs. Fig.5 Three scenarios of raiation pattern for 2 2- beamformer when an i stan for esire an interference signals, respectively. 6 Experimental Results To confirm the beamforming capability of the propose concept, a prototype of the beamformer is constructe, which is constitute by three major parts: array antennas, beamforming network an null locating network. For the array antennas, 4 stanar monopole antennas are employe in which it provies gain of 5 Bi iniviually. The array is arrange in 2 2 lattice. The array spacing is halfwavelength at 2.45 GHz. For beamforming network, the moifie Butler matrix presente in [8] is aopte here. It is aopte which provies 4 input ports an 4 output ports. The photograph of utilize beamforming network is shown in Fig. 6. The 4 input ports are connecte to the array antennas while the 4 output ports eliver the receive signals to null locating network. The 4 output signals from the beamforming network are the signals coming from 45, 135, 225 an 315 off boresight irection as well as for transmitting moe. Next the null locating network will provie the main beam to esire irection an null to interference irections as explaine earlier. The prototype of null locating networks are constructe an teste. The irections of interest are given at 45, 135, 225, 315. If one irection is chosen to be esire irection, the rest irections become interference irections. Fig.7 shows a prototype of null locating networks which has to be connecte to the beamformer presente in Fig.6. Top-left, top-right, bottom-left an bottom-right networks present the null locators for pointing main beam at 45, 135, 225 an 315, respectively, while keeping the other irections to nulls. The with an length of strip shown in Fig.7 represent amplitue an phase for interference suppressing ISSN: Issue 12, Volume 9, December 2010

5 Table 1. Interference Suppressing Coefficients measure from prototype show in Fig. 7. Coefficient(k q ) k 1 k 2 k 3 Main beam i i i i i i i i i i i i Fig.6 Moifie Butler matrix having the main beam irecte to 45, 135, 225 an 315. coefficients calculate in (5). Table 1 shows the measure outputs from the constructe prototype of null steering network shown in Fig.7. Next, the performance of constructe prototype is teste an one of four configurations shown in Fig.8 to 11 whereas ash line represents raiation pattern of the array without null locating an soli line inicates the one when incluing null locating. To valiate the use of propose system for WMNs, the measurement is require to be unertaken. The basic configuration of WMNs is illustrate in Fig.12. In this figure, five noes represente mesh routers are employe an the propose switche-beam is implemente at the center noe. Accoring to non-soli stanar for WMNs, one approach of air interface is able to be arrange by using conventional IEEE a/b/g. In this paper, four WLAN access points are locate at the corner as shown in Fig.12. For the propose switche-beam, the illustration of measurement setup at the center noe is epicte in Fig.13. To measure signal strength, there is no compensation ue to all power loss cause by connectors, Fig.7 Photographs of null locators. transmission lines an combiner. Hence, it makes sure that the propose system is practically teste for real application an promptly use for WMNs as its presente form. Figs. 14 an 15 present the receive signal strength versus power transmission for conventional omni-irectional antenna an propose switche-beam antennas. Each time step is a consecutive one-minute interval. For omni-irectional antenna, the receive signal strength of four access points varies from -38 to -70 Bm an there is no ominant access point. In turn, the results of propose switche-beam antennas provie two significant groups which are the access point in esire irection, Ap1, an null irections, Ap2 to Ap4. The gap between signal strength of esire access point an the others is range from 6 to 20 B. It is interesting to observe that the receive signal strength of esire access point varies from -33 to -35 Bm which is higher than the best signal strength of omni-irectional antenna. These results confirm the use of propose switchebeam antennas to enhance signal strength as well as suppress interference signals. ISSN: Issue 12, Volume 9, December 2010

6 Fig.8 Raiation pattern of a 2 2-beamformer when the esire signal is coming from 45 an interference signals are coming from 135, 225 an 315. Fig.9 Raiation pattern of a 2 2-beamformer when the esire signal is coming from 135 an interference signals are coming from 45, 225 an 315. Fig.10 Raiation pattern of a 2 2-beamformer when the esire signal is coming from 225 an interference signals are coming from 45, 135 an 315. Only signal strengths cannot totally inicate the performance of WMN so the next task is to valiate the other parameters to mark a merit of networks incluing elay an throughput. Fig.16 provies the average elay versus ata size when ownloaing. As seen in the figure, propose switche beam antennas can help the system to ownloa more quickly for both 2 an 6 Mbytes. This is because the propose system having higher receive signal strength experiences a lower outage probability than omni-irectional antenna. Also the signals from esire access point have a lower interruption interfere by other access points. Fig.17 presents the probability ensity function of throughput. In this figure, the total 100 samples are measure an evaluate through well known website, Fig.11 Raiation pattern of a 2 2-beamformer when the esire signal is coming from 315 an interference signals are coming from 45, 135 an 225. com.the results show that the omni-irectional antenna provies lower mean throughput than switche beam antennas an the probability having throughput more than mean value of omniirectional antenna is less than switche beam antennas. These results confirm the use of propose switche beam to enhance signal quality as well as suppress interference signals. In practice, the position of mesh router is not always on beam irection. Therefore, the following measurements base on configuration in Fig.18 are unertaken to investigate whether the propose system can provie benefits or not if mesh location is shifte from beam irection. Fig.19 presents the. receive signal strength versus angles eviation for omni-irectional antenna an propose switche- ISSN: Issue 12, Volume 9, December 2010

7 beam antennas. We can see when angle is eviate +15 an -15, switche-beam antennas still provie a receive signal strength of esire access point better than omni-irectional antenna. Fig. 10 provies the average elay versus ata size when ownloaing. As seen in the Fig. 20, the propose switche-beam antennas can help the system to ownloa more quickly for both 2 an 6 Mbytes. Fig.21 presents the probability ensity function of throughput. In this figure, the total 100 samples are measure an evaluate through well known website, The results show that the omni-irectional antenna provies lower mean throughput than switche-beam antennas an the probability having throughput higher than mean value of omni-irectional antenna is less than switche-beam antennas of all angle eviation. These results inicate the success of propose system even the mesh router is not installe on the irection of main beam. Signal strength(bm) Ap1 Ap2 Ap3 Ap Time Fig.14 Receive signal strengths from four access points by using conventional omni-irectional antenna Signal strength(bm) Fig.12 Scenarios for testing the propose system. Ap1-80 Ap2 Ap3 Ap Time Fig.15 Receive signal strengths from four access points by using switche-beam with null locating networks Omni-irectional Switche beam 25 Average elay (sec ) Fig.13 Measurement setup of propose system for WMNs Downloa size (Mbyte) Fig.16 Average elay when ownloaing ata via conventional omni-irectional antenna an propose switche-beam antennas. ISSN: Issue 12, Volume 9, December 2010

8 Probability ensity function Omni-irectional Switche beam Average elay(sec) Omni-irectional Switch beam (+15 Degree) Switch beam (Center) Switch beam (-15 Degree) Throughput (kbps) Fig.17 Probability ensity function of measure throughput when using conventional omniirectional antenna an propose switche-beam antennas Downloa size (Mbyte) Fig.20 Average elay when ownloaing ata for conventional omni-irectional antenna an propose switche-beam antennas Omni-irectional Switch beam (+15 Degree) Switch beam (Center) Switch beam (-15 Degree) Probability ensity function Fig.18 Test scenarios on location shift of mesh router Throughput (kbps) Fig.21 Probability ensity function of measure throughput when using conventional omniirectional antenna an propose switche-beam antennas. Fig.19 Receive signal strengths vs. angle eviation for four access points by using conventional omniirectional an propose switche-beam antennas. 7 Conclusion This paper has presente a low profile beamformer suitable for WMNs. Also, this beamformer is able to suppress interference signals by locating nulls to interfering irections. This beamformer is constitute by 2 2 antenna array space by λ/4 accompanie with a simple beamforming network. The null locating algorithm can be hanle by any economic microprocessor as it is only multiplication to constant values. The simulation results confirm its beamforming capability. Moreover, the experimental results have shown that the beamformer can improve the overall signal strength, throughput an elay over the use of conventional ISSN: Issue 12, Volume 9, December 2010

9 omni-irectional antenna. The prototype offer benefits in term of signal quality, throughput an elays. Also its benefits can be achieve even the mesh router is locate by + 15 egree eviating from beam irection. Acknowlegments Authors woul like to acknowlege financial support from Suranaree University of Technology, Thailan. Reference: [1] B.P. Crow, I. Wijaja, L.G. Kim, P.T. Saka IEEE Wireless Local Area Networks, Communications Magazine, IEEE, Vol.35, 1997, pp [2] Ian F. Akyiliz, Xuong Wang, an Weilin Wang, Wireless Mesh Networks: A Survey, Computer Networks an ISDN Systems, Vol. 47, 2005, pp [3] F. Babich, M. Comisso an L. Mania, Multi- Antenna Techniques for Wireless Mesh Networks in an Outoor Environment, Communications ICC '07 IEEE International Conference, 2007, pp [4] S. P. Applebaum, Aaptive Arrays, IEEE transaction on Antennas an Propagation, Vol.24, No.5, 1976, pp [5] W.R. L C.Y. Chu, K.H. Lin, S.F. Chang, Switche-Beam Antenna Base on Moifie Butler Matrix with Low Sielobe Level, Electronics Letters, Vol.40, 2004, pp [6] B. Raman an K. Chebrolu. Design an Evaluation of a New MAC for Long-Distance Mesh Networks, In The 11th Intl. Conference on Mobile Computing an Networking (MOBICOM), Colongne, [7] M. Uthansakul an P. Uthansakul, Low Profile DOA Finer at 2.4 GHz, Microwave an Optical Technology Letter, to be publishe on January, [8] H.M. Elkamchouch M.A.R.M. Aam, A New Constraine Fast Null Steering Algorithm, Antennas an Propagation Society International Symposium IEEE, Vol.2, 2000, pp [9] H. Steyskal, R.A. Shore an R.L. Haupt, Methos for Null Control an Their Effects on the Raiation Pattern, IEEE Trans. Antennas an Propagation, Vol.AP-34, 1986, pp [10] J.A. Stine, Exploiting smart antennas in wireless mesh networks using contention access, IEEE Wireless Communications, Vol.13, No.2, 2006, pp [11] V. Jain, A. Gupta, D.P. Agrawal, On-Deman Meium Access in Multihop Wireless Networks with Multiple Beam Smart Antennas, IEEE Transactions on Parallel an Distribute Systems, Vol.19, No.4, 2008, pp [12] N. Sangmanee, P. Uthansakul, R. Wongsan, M. Uthansakul, Fining capability enhancement for 360-egree DOA finer, ECTI-CON 2009, Vol.2, 2009, pp [13] E. Boyer, A. Ferreol, P. Larzabal, Simple robust bearing-range source's localization with curve wavefronts, IEEE Signal Processing Letters, Vol.12, No.6, 2005, pp [14] D. Astely, B. Ottersten, The effects of local scattering on irection of arrival estimation with MUSIC, IEEE Transactions on Signal Processing, Vol.47, No.12, 1999, pp Suwanna Praittara receive B.Eng egree from Suranaree University of Technology, Thailan, in 2008 an she is pursuing her M.Eng egree school of Telecommunication Engineering, Faculty of Engineering, Suranaree University of Technology, Thailan. Her current research interests inclue smart antenna, wireless mesh network. Monthippa Uthansakul receive B.Eng egree from Suranaree University of Technology, Thailan, in 1997 an M.Eng egrees from Chulalongkorn University, Thailan in She has joine Suranaree University of Technology since During , she stuie PhD at University of ueenslan, Australia, in the area of smart antenna especially wieban beamforming. She currently works as Assistant Professor in school of Telecommunication Engineering, Faculty of Engineering, Suranaree Unviersity of Technology, Thailan. She wrote 1 book chapter entitle Wieban smart antenna avoiing tappe-elay lines an filters in Hanbook on Avancements in Smart Antenna Technologies for Wireless Networks, Iea Group Publishing, USA, 2008 an she has publishe more than 50 referee journal an conference papers. Her current research interests inclue antenna array processing, compact switche-beam antenna an boy communications. Dr. Uthansakul receive Young Scientist Contest 2n Prize at 16th International Conference on Microwaves, Raar an Wireless Communications, Krakow, Polan, May ISSN: Issue 12, Volume 9, December 2010

10 Peerapong Uthansakul (M 09) receive B.Eng an M.Eng egrees from Chulalongkorn University, Thailan in 1996 an 1998, respectively. In , he worke as Telecommunication Engineer with Telephone Organization of Thailan (TOT) an then he has joine Suranaree University of Technology since During , he stuie PhD at University of ueenslan, Australia, in the area of wireless communications especially MIMO technology. He currently works as Assistant Professor in school of Telecommunication Engineering, Faculty of Engineering, Suranaree Unviersity of Technology, Thailan. He wrote 1 book entitle Aaptive MIMO Systems: Explorations for Inoor Wireless Communications (also available on amazon.com) an he has publishe more than 60 referee journal an conference papers. His current research interests inclue MIMO, OFDM, WiMAX, Diversity an Wireless Mesh Network. Dr. Uthansakul receive 2005 Best Stuent Presentation Prize winner at the 9th Australian Symposium on Antennas, Syney, February 2005, Australia an 2004 Young Scientist Travel Grant winner at the 2004 International Symposium on Antenna an Propagation, August 2004, Japan. ISSN: Issue 12, Volume 9, December 2010