Design of APRS Network using Low Cost Nanosatellite

Transcription

1 Design of APRS etwork sing Low Cost anosatellite Adnane ADDAIM, Abdelhak KHERRAS, El Bachir ZAOU, Abdelhafid ER-RADI Centre de Recherches et d Etdes Spatiales Ecole Mohammedia d Ingéniers, Rabat, Morocco. University of Mohammed V-Agdal {addaim, kherras, zanto, er-radi}@ emi.ac.ma Contact details of the first athor: Adnane ADDAIM Centre de Recherches et d Etdes Spatiales Ecole Mohammadia d Ingéniers, Rabat, Morocco. University of Mohammed V-Agdal el.: Fax.: addaim@emi.ac.ma Abstract he se of anosatellite to provide APRS network service is discssed. he satellite, as small as the cbesat concept, reqires employment of a very simple payload whereas all the complexity is broght back to the terminals. he radio link between the terminals and the anosatellite will be stdied. We will employ the Slotted-Aloha in the plink, and establish a mathematical model for evalating the performance of the system nder finite nmber of terminals. In sch a system, two appropriate measres of the performance are the throghpt and the average delay; ths, we will se these measres in this paper. Keywords anosatellite, Slotted ALOHA, Link Bdget, APRS, Satellite Commnication.

2 Design of APRS etwork sing Low Cost anosatellite A. Addaim, A. Kherras, B. Zanto, A. Er-radi a Centre de Recherches et d Etdes Spatiales Ecole Mohammadia d Ingéniers University of Mohammed V-Agdal, Rabat, Morocco. Abstract he se of anosatellite to provide APRS network service is discssed. he satellite, as small as the cbesat concept, reqires employment of a very simple payload whereas all the complexity is broght back to the terminals. he radio link between the terminals and the anosatellite will be stdied. We will employ the Slotted- Aloha in the plink, and establish a mathematical model for evalating the performance of the system nder finite nmber of terminals. In sch a system, two appropriate measres of the performance are the throghpt and the average delay; ths, we will se these measres in this paper. Keywords anosatellite, Slotted ALOHA, Link Bdget, APRS, Satellite Commnication. I. IRODUCIO HE Cbesat project [] aims to send a satellite, with dimensions 0x0x0 cm and mass one kilogram into Low Earth Orbit, designed and bilt by stdents. he mission of the anosatellite is to provide varios APRS services, sch as mobile localization, messaging and data collection sing a store-and-forward payload. he APRS stands for Atomatic Position Reporting System. which was developed by Bob Brning [2]. A central station cold gather data via the anosatellite from world wide remote sites, where hman presence is expensive and difficlt to spport. hen the gathered data will be archived to a database and accessed by Internet. he PCsat system [2] demonstrates the power of sing internet linked amater satellite grond stations arond the world to captre live APRS telemetry data for se anywhere. II. SYSEM DESCRIPIO Consider a commnication network comprising Store and forw-ard Cbesat and a finite nmber of ser terminals. In the satellite system model, we assme a satellite in circlar orbit at the altitde h = 650 km with a minimm elevation angle E = 0. We assme that a total nmber of terminals are distribted randomly in a given service area arond the sbsatellite line. he maximm time of the visibility for each terminal is given by: /2 3 a R = 2 Arcos cos E µ a ( E). () R = km, and µ = 3, m 3 s -2. Where R is the effective Earth radis, µ is the constant of the gravitation and a is the semi-major axis of the satellite orbit with a= h + R. he maximm time shift between two terminals in the coverage area is given by: = D. (2) sh max V sat with, D max the maximm distance between two terminals in the coverage area is given by: D max anosatellite Data Collection Messaging Mobile Localization R = 2 R Arcos cos E a ( E). (3) For circlar orbit, V sat, the speed of the satellite, is constant [3] and is given by: ( a) 2 V sat = µ. (4) From (), (2), (3) and (4), we have: R a erminals sh =. (5) Radio nit Processing nit GPS Circlar Orbit 650 Km Store & Forward service area sbsatellite line Microcontroller Internet Central Station Fig.. APRS servives sing Store and forword anosatellite ransmitter Receiver Modem Display + Keyset User interface nit Memory Fig. 2. Fnctional Architectre of erminals VHF Antenna

3 he terminals [4] are eqipped with a processing nit which se an orbitography software to recognize the moments of the satellite passage and carries ot the Doppler correction of the transmitted and received freqencies. o make its calclations, the orbitography software needs three parameters to know, terminal geographical position, UC time and the ORAD- Keplerian Elements. he first two parameters are given by internal GPS receiver while the third parameter is given by the central station which reglarly commnicates ( to 2 times per abot fifteen days) the pdated ORAD-Keplerian Elements of the satellite to all the terminals. he fnctional architectre of the terminals is given by Fig. 2. A block diagram of the proposed APRS payload is shown in fig.3. he prpose of the On Board Data Handling (OBDH) is to extract information ot of APRSpackets, filter ot specific APRS-ID s and store them. he stored packets are to be formatted for download when the right telecommand is given from grond segment. III. RADIO LIK At VHF freqencies the atmosphere and ionosphere have little effect on the propagating radio wave [5]. Also, antennas, receivers and transmitters for both the grond and the space segment are readily available and inexpensive. Antenna gain ransmitted power Antenna Feed Loss syst ABLE I BUDGE LIK erminal parameters 0 dbi 5 W 0.5 db 2000 k Channel parameters Free Space Loss - 4 db * Polarization Losses Additional Losses Antenna gain ransmitted power Antenna Feed Loss syst APRS OBDH Bad Rate Modem ransmitter Receiver Fig. 3. Architectre of APRS Payload Satellite parameters Operational parameters VHF Antenna 3 db 0.5 db 0 dbi W 0.5 db 5000 k 200 bad E b / 0 reqired 3.6 db *2 Link margin for plink E b / 0 estimated 2.42 db *3 Margin 7.82 db Link margin for downlink E b / 0 estimated 8.42 db *3 Margin 4.82 db * Free space loss, FSL is calclated by [6]: db ( D ) 20log ( F ) FSL = log +. (6) km MHz where F MHz is the transmit freqency, and Dkm is the distance or range of the satellite, calclated by: D km 2 2 ( R + h) ( R.cos( E) ) R. sin( E) =. (7) *2 Reqired Eb/o gives an operating bit error rate of 2x0-5 [7], assming non-coherent demodlation of FSK. *3 An estimation of the Eb/o is obtained from: Eb P. Li. Gt. Ls. La. Gr =. (8) 0 k. s. R Where: P = ransmitted Power L i = Feed Losses G t = ransmit antenna gain L s = Free Space Loss (FSL) L a = x Path Losses (Miscellaneos) G r = Receive antenna gain K = Boltzmans constant s = System temperatre R = Data rate of the system he link bdget, shown in able I, demonstrates how an acceptable service can theoretically be obtained between a anosatellite, with simple omnidirectional antenna, orbiting in a 650 km low earth orbit operating at 45 MHz, and qarter wavelength antenna based terminal. If one looks at the cbesat projects they generally se abot one watt of otpt power while limiting the data rate to 200 bad. Since the central station Yagi antenna will be tracking the satellite, it will always be pointing in sch a way that the maximm gain, in or case 3 db gain, is achieved. herefore the margin between the anosatellite and the central station will be improved by 3 db. he FSK modlation scheme was chosen becase of its simplicity and proven robstness. Used in conjnction with an ARQ protocol the scheme provides error free commnications between the terminals and the anosatellite. IV. HROUGHPU AALYSIS In this system, the plink mltiple access from terminals to satellite is the limiting direction which is established according to slotted aloha. erminals start the transmissions of their informations in the form of the fixed-length packets at common clock instances whenever they have information to send. Slot size and packet length are assmed to be eqal. In conventional slotted Aloha [8], the throghpt can be defined as the expected nmber of sccessflly packets in each time slot, with the dimension of packets per slot. A transmitted packet can be received incorrectly or lost completely becase of two different types of errors:

4 ) random noise errors and 2) errors cased by packet collision. In this paper, we assme that the first type of error can be ignored, and we shall be concerned only by errors cased by packet collision. In this system, however, a terminal can always find ot whether a packet was destroyed by monitoring the timer. If the packet was involved a collision and destroyed, the terminal waits for random delay and then retransmits the packet. Let G i be the probability that the ith terminal will send a packet, inclding newly generated and retransmitted packets in each time slot and S i the probability that the ith terminal will transmit a packet sccessflly in each time slot. If all terminals are identical, i.e. G i = G for each i ={,.., }, so the probability that the nmber of the transmitted packets by terminals, dring τ, is eqal to k, follows the binomial law [9] with parameters G and K as: G ) k k Ck G k P ( k; G ) =. (9)..( 0 k > ote that G becase the terminal can not transmit more than one packet in a time slot. he probability of sccess P sc that terminal transmits sccessflly a packet, is eqal to the probability that ( -) other terminals don t send any packet dring time slot to avoid the collision, which corresponds to the probability P (0 ; G ) : G ) P = P(0; G ) =. (0) sc he probability of a sccessfl transmission S by the terminal is eqal to the probability G that the terminal sends a packet mltiplied by the probability of sccess P sc. G ) S = G. () We define the normalised channel hroghpt S and the normalised raffic Load G for the Slotted channel as : S = S i i= and G = G i i=. (2) Since all terminals are identical, we have S = S/ and G = G/. So () can be written ) S. (3) = G G ds/dg=0 for G =. So the maximm throghpt S max is: S max =. (4) he relation between the offered traffic load and the throghpt of the slotted Aloha scheme of (3) is plotted in Fig. 4, which shows that the maximm throghpt S max depends little on the nmber of terminals, and the throghpt S depend almost only on G, which is intitively reasonable becase, 0 terminals transmitting 4 packets have the same traffic load G as 40 terminals each one transmitting one packet. As can be seen in the Fig. 4, the peak of the throghpt for the slotted Aloha occrs at G=. hat means that if the system is operating at G=, the probabilty of empty slot is /e, or abot Operating at higher traffic loads redces the nmber of empties bt increases the nmber of collisions exponentially. We cold define the normalised G as G t + =. (5) where, t is the total nmber of transmitted packets (new and retransmitted packets) dring the time of visibility of the satellite, β is the nmber of slots available dring, which eqals /τ, and ω is the nmber of slots available with the consideration of the time shift between two extremities of the service area, ω = sh / τ. We cold define the normalised throghpt as: S sc + =. (6) where, sc is the nmber of sccessflly transmitted packets dring the time of visibility of the satellite. From (3), (5) and (6) we have: Fig. 4. hroghpt vs raffic Load for a Slotted ALOHA t sc = t. (7) + β=286 ; ω=259 β =46 ; ω=48 β=636 ; ω=577 Fig. 5. otal mber of packets vs nmber of sccessfl packets

5 λ Fig. 6. raffic Load vs Departre Rate for Slotted ALOHA Let λ be the probability that a terminal generates new packet in each time slot and the time of the visibility of the satellite by each terminal. In the model presented in this paper, it is assmed that the terminals have the same traffic reqirements. Each terminal transmits maximm one packet dring the visibility period of the satellite. We have λ expressed by: λ = +. (8) he condition of the stability [0] states that the departre rate λ (new packets rate) is eqal to the arrival rate S (hroghpt). With λ < S max, we have: G λ = G. (9) In the stable case where λ < S max and the retransmission procedre is random, the average of the probability of packet sccess P sc dring the visibility period of the satellite is given by: λ =. (20) P sc G with (G,λ) stable eqilibrim point. In the nstable case where λ > S max the offered traffic G increases qickly, every terminal retransmits. he average probability of sccess P sc reaches qickly to zero. According to (8), the response average delay D increases rapidly to infinite. V. AVERAGE DELAY AALYSIS In this section, we evalate the average delay performance of the Slotted Aloha. he packet can be generated at any point dring a slot, yet the terminal has to wait ntil the beginning of the next slot before attempting a transmission. he time between packet generation and the start of the next slot can have any vale between 0 and τ, the packet dration, with eqal probability. Hence, the average time from the generation of the packet ntil the entire packet enters the channel eqals 3/2 τ. he packet dration τ is given by τ = L/R with L the length of the packet and R the bit rate. In this paper, we will se R=200 bps and AX.25 protocol [] over the satellite link as sed by APRS. All APRS transmissions se AX.25 UI-frames with 9 fields of data. he length of the information packet in AX.25 protocol depends on the length of the information field by L = (27+ ) bytes, with maximm eqals to 256 bytes. As any network protocol, in AX.25 the 27 bytes is reserved to the overhead, which contains the addresses of the both terminals involved in the same commnication and the satellite, and other fields related to the control of the commnication. For the efficiency of the protocol we will keep the percentage of the overhead less than 20% of the whole packet, hence, the length takes 00 octets in his minimm. herefore, the packet dration τ is between 0.85 and.89 seconds. For the sake of simplicity in calclations, Zero-gard time is assmed here. he packet travels throgh the plink satellite channel and is sbjected to the one-hop satellite delay, d. In the case of sccessfl reception of the packet by the satellite, on average = 3/2 τ + d seconds elapse between generation of the packet and its fll acceptance by satellite. In this case, the satellite sends an acknowledgment packet with dration of ack, and the ser receives that packet after d + ack + proc seconds, where proc is the reqired satellite processing time. From the moment of sending packet, the ser starts a timer of dration t = d + τ + proc + ack + d, expecting to receive the acknowledgment packet. If the attempt is sccessfl, the ser clears the packet. If an acknowledgment packet is not received, the ser considers the packet lost and starts the process of reattempting transmission after expiring of his timer t. In the satellite system with altitde h= 650 Km, d is abot 4 ms, hence, it is negligible compared to packet dration τ. Here we assme that ack and proc are negligible compared to packet dration. It is reasonable to assme that the probability of loss of the acknowledgment packet is very small. o optimize the policy of retransmission by avoiding collisions of the same packets, the lost packet is retransmitted at the end of a random time chosen niformly in the interval [0τ, δ max τ] with the average δ moy = δ max /2 (δ max is an integer) [2]. We assme that δ max is eqal to the nmber of ser to randomize the retransmission procedre. he average delay of the packet dring the visibility of the satellite can be expressed as [3,4]: αmax ( sc + + kw k= )( P ) k sc D = P P. (8) where w is the average time elapsed between the end of the first transmitted packet and the end of the first retransmitted packet or between the ends of any two sccessive retransmitted packet, at the satellite, eqals to w = t + δ moy τ. And ( - P sc ) k P sc is the probability that a packet being received in the (k+)th attempt of transmissions after k times of failres retransmissions, with P sc is the average of the probability of packet sccess dring the visibility period of the satellite. sc

6 β=636 ; ω=577 β =46 ; ω=48 β=286 ; ω=259 Fig. 7. Average Delay vs mber of erminals [8]. Abramson, " Fndamentals of Packet Mltiple Access for Satellite etworks," IEEE J. Select. Area Commn., Vol. 0, o. 2, pp , 992. [9] L. Kleinrock, «Qeing Systems» Vols. and 2, ew York : Wiley, 975. [0] J. Massey, "Some new approaches to random access commnications," IEEE press, Mltiple Access Commnications, pp , 992. [] APR, "AX.25 Link Access Protocol for Amater Packet Radio," APR, 2.2 edn., 997. [2] F. obagi, M. Gerla, R. Peebles, and E. Manning, " Modeling and Measrement echniqes in Packet Commnication etworks Proceedings of the IEEE, Vol. 66, o., 978. [3] A. Jamalipor, «Low Earth Orbital Satellites for Personal Commnication etworks», orwood, MA: Arthech Hose, 998. [4] H.ab, D. Schilling, «Principles of commnication systems» McGraw-Hill, 2 ème Edition 987. And + k w is the time elapsed between the generation of the packet and his fll acceptance after (+k)th attempt. Becase t is smaller than 2τ and the terminal has to start at the beginning of the next slot, in this case, w becomes eqal to (2 +δmoy)τ. he maximm nmber of attempt α max and the maximm waiting delay δ max τ are limited by the time of visibility of the satellite available for each terminal with + α max (2+δmax)τ <. VI. COCLUSIO In general, satellite will not compete with terrestrial data networks. he niqe performance capability reslting from the se of the Cbesat spports the provision of APRS service to developing contries and rral areas where there is no wired infrastrctre of commnication. Besides being more economic to the rral areas, the satellite network are not affected by the natral disasters and contine to give the service in sch sitations. In this research, we find that the throghpt reaches its maximm bt the only drawback is a long average delay which is acceptable for many applications which se a single LEO satellite. hroghpt and average delay are very difficlt to figre ot by an analytical method in sch system. herefore, as a ftre work, reslts worked ot in this paper shold be validated by simlations sing a discrete event simlation tool. REFERECES [] S. Waydo, D. Henry, and M. Campbell, CbeSat Design for LEO- Based Earth Science Missions, IEEE Aerospace Conference Proceedings, Vol., pp , [2] Bob Brninga, A Description of APRS by WB4APR, Packet Users otebook, CQ Magazine, Dec 993. [3] M. Wirner, A. Jahn, E. Ltz and A. Bottcher, " Analysis of system Parameters for LEO/ICO Satellite Commnication etworks," IEEE J. Select. Areas Commn., vol 3, 2, 995. [4] B. Zanto and A. Kherras, "Design of atomatiqe balise por localisation par microsatellite," ISIVC 04, France, [5] J. Paffet,. G. Jeans, J. Ward, "VHF-Band Interfrence Avoidance for ext-generation Small Satellites," 2 th AIAA/USU Conference on Small Satellites, SSC98-II-5, 998. [6] G. Maral, M. Bosqet, «Satellite Commnication Systems» John Wiley & Sons; 4 edition, May, [7] J. Proakis, «Digital Commnications» McGraw-Hill, Second Edition, 989.