Dynamic Resource Allocation Algorithm of UAS by Network Environment and Data Requirement

Transcription

1 Dynamc Resource Allocaton Algorthm of UAS by Network Envronment and Data Requrement Hye-Rm Cheon, Jun-Woo Cho, and Jae-Hyun Km Department of Electrcal and Computer Engneerng Ajou Unversty Suwon, Republc of Korea {hyermn, cjw8945, Abstract Due to the expanson of the unmanned aeral vehcle (UAV) market, there s the ssue of spectrum scarcty for unmanned aeral system (UAS). Thus, t needs to allocate the resource effectvely n the lmted bandwdth consderng the network envronment. In ths paper, we propose the structure and the resource allocaton algorthm whch can maxmze the network throughput as well as satsfy the mnmum data rate requrement By performance analyss, we show that the proposed algorthm can allocate the resource to satsfy the hgh network throughput as well as the mnmum data requrement n the gven network envronment. Keywords UAV; UAS; TDMA; Resource Allocaton; Scheduler I. INTRODUCTION Durng the last few years, the growth of unmanned aeral vehcles (UAV) market shows a sgnfcant ncrease due to the varous applcatons n cvl and mltary areas and ts ncrease trend wll contnue n next few decades. In Teal Group s report, ts analysts forecast that worldwde mltary and cvl unmanned aeral systems (UAS) combned producton wll soar to $0.3 bllon n 05, up from $5.4 bllon n 06, and total $35 bllon over the next 0 years [, ]. In the expanson of UAS market, to relably operate the multple UAS and have the nteroperablty between varous UAS, t needs to make a standardzed specfcaton of UAS. Thus, there are a few standardzaton works for UAS, such as the mnmum operaton performance Standard (MOPS) for UAS by Rado Techncal Commsson for Aeronautcs (RTCA), standardzaton agreement (STANAG) 4586 by North Atlantc Treaty Organzaton (NATO) [3, 4]. In addton, there have been studes on bandwdth requrements for UAS. In Radocommuncaton Sector of Internatonal Telecommuncatons Unon (ITU-R) s report, they estmate that the maxmum amount of bandwdth requred for UAS are 34 MHz and 56 MHz for satellte and terrestral systems respectvely [5]. In Natonal Aeronautcs and Space Admnstraton (NASA) s study, the bandwdth for UAS communcatons estmates 0.4 MHz for networked envronment, MHz for non-networked envronment [6]. Although the spectral requrement for UAS s derved as mentoned earler, there are the lack of suffcent bandwdth to allocate to the UAS communcatons lnks because not only the UAS market explosvely ncreases, but also the exstng manned aeral systems (MAV) are operated n current bandwdth, especally n the L-band [7]. In addton, the data types and requrements are dfferent dependng on the msson of the UAS. Furthermore, the envronment of UAS communcaton lnks, such as the number of UAVs, the weather and the communcaton channel condton s changed frequently because of the hgh moblty of UAV [8]. Thus, we propose the structure dynamcally choosng optmal unt tmeslot and resource allocaton algorthm whch can maxmze the network throughput as well as satsfy the mnmum data rate requrement for the UAS operaton n the gven network envronment based on Tme Dvson Multple Access (TDMA). The remander of ths paper s organzed as follows. In Secton II, we revew the relate works. In Secton III, we propose the dynamc resource allocaton algorthm based TDMA. In Secton IV, we analyze the performance of the proposed algorthm. Fnally, we conclude ths paper n Secton V. II. RELATED WORK Before the presentng the proposed algorthm, we study the related works of two categores: TDMA-based UAS MAC protocols and other multple access-based UAS MAC protocols. A. TDMA-based MAC protocols for UAS There are a lot of researches based on the TDMA MAC protocol. Jang et al. proposed a locaton-based TDMA MAC protocol to solve the transmsson delay problems where t uses multple UAVs envronment [9]. Jang et al. proposed a new TDMA MAC protocol for the transmsson delay problems. The proposed MAC protocol s used only one guard tme n a usng Pggy-backng algorthm [0]. Young et al. proposed a unfyng slot assgnment protocol (USAP) usng dstance vertex-colorng problem n slot segments dvded nto Bootstrap, Broadcast and Reservaton to avod nterference from -hop neghbor nodes []. Youns et al. proposed a cogntve USAP (C-USAP) for tme slot schedulng and Ths work has been supported by the Future Combat System Network Technology Research Center program of Defense Acquston Program Admnstraton and Agency for Defense Development (UD60070BD) /7/$ IEEE 384 ICTC 07

2 channel allocaton at the same tme usng dstrbuted method. A proposed protocol n [] s used rado nterference model []. B. Other Multple Access-based MAC protocol for UAS There are a lot of researches based on other multple access-based MAC protocol. Ho et al. proposed a based random access (FRA) method usng prortzed selecton (PFS) method to reduce packet error rate (PER) between UAS and wreless sensor networks [3]. Temel at al. proposed a locaton orented drectonal MAC (LODMAC) protocol usng drectonal antennas based CSMA/CA. A proposed protocol s used Busy to Send (BTS) packet along wth the Request to Send (RTS) and Clear to Send (CTS) to overcome drectonal deafness problem [4]. Gu et al. proposed a centralzed ntellgent channel assgned multple access (C-ICAMA) combned Slotted ALOHA and TDMA for ground node to access UAV to solve the asymmetrc traffc. A proposed method can dynamcally allocate bandwdth for uplnk and downlnk usng ntellgent schedulng algorthm [5]. III. PROPOSED RESOURCE ALLOCATION ALGORITHM We propose the dynamc resource allocaton algorthm whch determne the tmeslot allocaton for unmanned aeral system (UAS) based on the TDMA. In the algorthm, we determne the optmal unt tmeslot sze wthout tmeslot allocaton waste and then allocate the tmeslot for each data type to maxmze the network throughput under gven network condton and data rate requrement. Frst, we propose the structure for ths algorthm as shown n Fg.. In ths paper, we decde the sze as 00 ms because the maxmum message update rate for UAS s 0 Hz.e. t should update messages at least once per 00 ms [6]. Then, a ncludes a number of tmeslots for crtcal data and uncrtcal data, and each tmeslot conssts of a few unt tmeslots and guard tme. The sze of guard tme defnes as about 5% of the unt tmeslot sze, based on [6]. The proposed resource allocaton algorthm s descrbed n Algorthm. The algorthm has 8 phases. In phase, we calculate the requrement data volume for each data type per a, whch means that a UAV should transmt the data volume wthn a to satsfy the requrement data rate and t s gven by D R T I (), where D s the requrement data volume sze for data type, R s the requrement data rate of data type, T s the sze, whch s defned as 00 ms n ths paper. Data type means one of data types n set I and ths data type set ncludes crtcal data types to operate the UAS such as uplnk message and downlnk message, n addton, the varous other data types such as voce, vdeo, and so on accordng the msson of UAS. Next, n phase, we determne the optmal unt tmeslot sze. Frst, we choose the ntal unt tmeslot sze T unt_fxed, whch should be much smaller than a sze consderng the number of data types. Next, we calculate the requrement tme for each data type to transmt requrement data volume D wthn a and t s defned as t req _ D. () spectral effcency bandwdth And then, we defne the unt tmeslot sze for data type, T unt_ that t rounds t req_ up to the frst dgt place of t req_, whch descrbes n lne 5- of algorthm. Fnally, we choose the mnmum value of T unt_fxed and T unt_ as unt tmeslot sze, T unt_. After decde the unt tmeslot sze, n phase 3, we calculate the unt data volume per a unt tmeslot, whch mean that t can transmt the data volume wthn a unt tmeslot under the network envronment and t s gven by D unt spectral effcency bandwdth. (3) T unt Crtcal Data Crtcal Data Crtcal Data # Unt Unt Guard Tmeslot Tmeslot Tme Fg.. Frame structure Unt Tmeslot * cel(d /D unt ) + Guard Tme Uncrtcal Data Crtcal Data # Uncrtcal Data In phase 4, we calculate the allocaton requrement tme for each data type per a, whch means that t needs to allocate the tme to transmt the data wthn a to satsfy the requrement data rate and t s gven by D t Tunt Tguard, I, (4) Dunt where t s the allocaton requrement tme for data type per a, T guard s guard tme sze and s defned as a quarter of T unt. x s celng functon, whch s the least nteger greater than or equal to x. 385

3 Algorthm DTDMA UAV Resource Allocaton Algorthm : Phase Calculate the requrement data volume per : D, RT I 3: Phase Determne the unt tmeslot sze 4: D treq _ spectral effcency bandwdth 5: for = to I do 6: j = 0 7: whle t req _ 8: treq_ treq_ 0 9: j = j+ 0: end whle : T t : end for unt_ req _ 0 j 3: Tunt mn Tunt_fxed, Tunt _ I 4: Tguard Tunt 0.5 5: Phase 3 Calculate the unt data volume per unt tmeslot 6: spectral effcency bandwdth Dunt Tunt 7: Phase 4 Calculate the allocaton requrement tme per 8: D t Tunt Tguard, I Dunt 9: Phase 5 Calculate the maxmum number of UAVs to be acceptable for the crtcal data allocaton 0: T nmax // I s the set of crtcal data t I : Phase 6 Decde the number of tmeslot to allocate for crtcal data n 0, n, I : max 3: Phase 7 Decde the number of tmeslot to allocate for uncrtcal data 4: f T tnmn t I I // I I I I I, 5: n 0, I 6: else 7: Drev T tn I spectral effcencybandwdth 8: D n w D w arg max w 0, rev I I T 9: n w T t n t, I I 30: end f 3: Phase 8 Calculate the allocaton tme per 3: T t n, I In phase 5, we calculate the maxmum number of UAVs to be acceptable when t can only allocate the tmeslot for crtcal data types such as messages and t s defned as n max T t I, (5) where n max the maxmum number of acceptable UAVS and I s the set of crtcal data type. The crtcal data type means that t s postvely necessary to operate the UAS and ths data types vary accordng to the purpose of UAS. x s floor functon, whch s the greatest nteger less than or equal to x. In phase 6, we decde the number of tmeslot to allocate for each crtcal data type, whch s not larger than n max. In phase 7, we decde the number of tmeslot to allocate for other data types,.e. uncrtcal data type. Frst, we check the allocaton avalablty durng the remanng tmeslot after the tmeslot allocaton for crtcal data type as follows, mn T t n t I, (6) I where, n s the number of tmeslot for data type, I s the set of uncrtcal data type. The unon of I and I s I and the ntersecton of I and I s empty set. Eq. (6) means that t cannot allocate the tmeslot for uncrtcal data when remanng tme s smaller than the mnmum value of the allocaton requrement tme for uncrtcal data type. When satsfy the eq. (6), we cannot allocate the tmeslot for uncrtcal data type. On the other hand, we can allocate. In ths case, we allocate the tmeslot accordng to the weghtng factor of each data type to maxmze the network throughput. To fnd ths weghtng factor, we calculate the data volume whch s able to transmt durng the remanng tme and t s gven by Drem T tn spectral effcencybandwdth (7) I where D rem s the transmsson-avalable data volume wthn the remanng tme. And then, we fnd the optmal weghtng factor set to maxmze the network throughput as follows, 386

4 D n w D rem I I w arg max, (8) w 0, T where w s the optmal weghtng factor set, w { w}, I, and total sum for the elements of w s. Based the optmal weghtng factor, we calculate the number of tmeslot to allocate for uncrtcal data and t s defned as n wt tn t, I. (9) I Fnally, n phase 8, we calculate the allocaton tme for each data type per and t s gven by T t n, I, (0) where T s the allocaton tme for data type. IV. PERFOMANCE ANALYSIS To evaluate the performance of the proposed algorthm, we perform the smulaton usng MATLAB. We compare the proposed algorthm wth the algorthm whch has ms fxed unt tmeslot sze. The performance analyss envronment refers to [7] and [6], and set them as Envronment and Envronment. The bandwdth sets MHz n Envronment and 6 MHz n Envronment. There are lower than the requred bandwdth of [7] and [6] n order to make the assumpton that the requred bandwdth s nsuffcent. The other parameters are lsted n Table. Table s the result of the maxmum number of UAVs to be acceptable for the crtcal data allocaton (n max) whch are resulted from Phase 5 n the Algorthm. The legend of Dynamc refers to the proposed resource allocaton algorthm and the legend of Fxed refers to the resource allocaton algorthm whch has ms fxed unt tmeslot sze. Table 3 s the result of the average data rate per UAV. The uncrtcal data (Voce, Vdeo) may not be allocated due to the optmal weghtng combnaton. So the values are averaged out only when data s allocated. We have confrmed that the values are above the requred data rate. In Fg., we show the total network throughput accordng to the number of UAVs n Envronment. In Envronment, the total network throughput s almost same n Dynamc and Fxed, because there s no redundancy tme n unt tmeslot. So, throughput may be ncreased by selectng the large value of unt tmeslot. In Fg. 3, we show the total network throughput accordng to the number of UAVs n Envronment. In Envronment, the total network throughput whch case of Dynamc s hgher than Fxed when the number of UAVs s ncreased, because the sze of the unt tmeslot s fxed largely so that data can be transmtted wthn a unt tmeslot due to the data rate of uplnk and voce s low n Envronment. That s, the redundancy tme occurs untl next unt tmeslot when use Fxed. Thus, the throughput can be ncreased by selectng the unt tmeslot for the requred data rate. TABLE II. THE MAXIMUM NUMBER OF UAVS Envronment Dynamc Fxed Envronment TABLE I. PERFORMANCE ANALYSIS ENVIRONMENT Envronment Parameter Envronment Envronment Modulaton QPSK OQPSK TABLE III. AVERAGE DATA RATE PER UAV Code rate Uncoded Spectral effcency Bandwdth (MHz) Requrement data rate Uplnk Downlnk k k Uplnk Downlnk 4,933 8,774 Voce 7.7 k Voce 4,800 Vdeo 70 k Vdeo M Parameter Dynamc Fxed Uplnk 38,435 38,409 Downlnk Envronment 50,460 50,373 Voce 34,633 34,546 Envronment Vdeo 8,50 8,55 Uplnk 5,63 6,8 Downlnk 3,384 35,96 Voce 5,30 5,986 Vdeo,003 k,003 k 387

5 Throughput.5 x 06 Network Throughput (Envronment ) Number of UAVs Fg.. Total network throughput n envronment Throughput x 06 Network Throughput (Envronment ) Dynamc Fxed Dynamc Fxed Number of UAVs Fg. 3. Total network throughput n envronment V. CONCLUSION We propose the new structure based on the dynamc selecton unt tmeslot and TDMA resource allocaton algorthm n gven network envronment, whch can maxmze network throughput whle satsfy the mnmum data rate requrement of UAS. In performance analyss, the network throughput of proposed algorthm s hgher than the algorthm usng fxed unt tmeslot whle satsfy the mnmum data rate requrement n gven network envronment. REFERENCES [] Systems Market Profle and Forecast," Oct. 06. [] Teal Group, "06 World Cvl Unmanned Aeral Vehcle Systems Market Profle and Forecast," Jul. 06. [3] RTCA, "SC-8 Mnmum Operatonal Performance Standards for Unmanned Arcraft Systems," [4] NATO Standardzaton Agency, "STANAG 4586 (Edton 3) - Standard Interfaces of UAV System (UCS) for NATO UAV Interoperablty," NSA/35(0)4586, Nov. 0. [5] ITU-R, "Characterstcs of Unmanned Arcraft Systems and Spectrum Requrements to Support Ther Safe Operaton n Non-Segregated Arspace," ITU-R M.7, Dec [6] S. Henrksen, "Unmanned Arcraft System and ATC Communcatons Bandwdth Requrements," NASA/CR , Feb [7] H. W. Km, K. S. Kang, D. I. Chang, and J. Y. Ahn, "Techncal and standardzaton trends on and non-payload communcatons for unmanned arcraft systems," Electroncs and Telecommuncatons Trends, vol. 30, no.3, pp.74-83, Jun. 05. [8] İ. Bekmezc, O. K. Sahngoz, and S. Temel, "Flyng ad-hoc networks (FANETs): a survey," Ad Hoc Networks, vol., no. 3, pp , Jun. 04. [9] H. Jang, E. Km, J. Lee, and J. Lm, Locaton-based TDMA MAC for relable aeronautcal communcaton, IEEE Transactons on Aerospace and Electronc Systems, vol. 48, no., pp , Apr. 0. [0] H. Jang, H. Noh, and J. Lm, "Arborne TDMA for Hgh Throughput and Fast Weather Condtons Notfcaton," Internatonal Journal of Computer Networks & Communcatons, vol.3, no.3, pp.06-0, May. 0. [] C. D. Young, "USAP Multple Access: Dynamc Resource Allocaton for Moble Mult-hop Multchannel Wreless Networkng," n Proc. IEEE MILCOM 999, November 999 [] O. Youns, D. Shallcross, L. Kant, K. Young, C. Graff, M. Patel, "TDMA Schedulng and Channel Assgnment for Cogntve Tactcal Networks," n Proc. IEEE MILCOM 0, Oct. 0. [3] D. T. Ho, and S. Shmamoto, "Hghly Relable Communcaton Protocol for WSNUAV System Employng TDMA and PFS Scheme," n Proc. IEEE GLOBECOM Workshops 0, Dec. 0. [4] S. Temel, and I. Bekmezc, LODMAC: locaton orented drectonal MAC protocol for FANETs, Comp. Net., vol. 83, pp.76-84, Jun. 05. [5] D. L. Gu. X. Hong, M. Gerla, G. Pe, and Y. Z. Lee, "C-ICAMA, a centralzed ntellgent channel assgned multple access for mult-layer ad-hoc wreless networks wth UAVs," n Proc. IEEE WCNC 000, pp , Chcago. IL, USA, Sep [6] H. Baek, J. Lm, J. Koo, J. Jn, P. Chun, and I. Oh, "Relable Dynamc TDMA Scheme wth new Packng method for Image Transmsson over Lnk-6," Journal of KICS, vol.37, no., pp , Nov. 0. [7] J. A. Kakar, "UAV Communcatons: Spectral Requrements, MAV and SUAV Channel Modelng, OFDM Waveform Parameters, Performance and Spectrum Management," Vrgna Tech Master Theses [6970], May 05 ACKNOWLEDGMENT Ths work has been supported by the Future Combat System Network Technology Research Center program of Defense Acquston Program Admnstraton and Agency for Defense Development (UD60070BD). 388