scieee Science in your language
[en] (orig)

GlSch: Observation Scheduler for the GLORIA Telescope Network

Abstract

This paper proposes the design and development of a scheduler for the GLORIA telescope network. This network, which main objective is to make astronomy closer to citizens in general, is formed by 18 telescopes spread over four continents and both hemispheres. Part of the management of this network is made by the network scheduler. It receives the observation requests made by the GLORIA users and then sends it to the most suitable telescope. A key module of the network scheduler is the telescope decision algorithm that makes possible to choose the best telescope, and thus avoiding offering an observation to a telescope that cannot execute it. This paper shows two different telescope decision algorithms: the first one is only based on weather forecast, meanwhile the second one uses fuzzy logic and information from each network telescope. Both algorithms were deployed in the GLORIA network. The achieved results coupled with a comparative of their performance is shown. Moreover, the network scheduler architecture, based on a hybrid distributed-centralized schema, is detailed.

Read accessible full text

GlSch: Observation Scheduler for the GLORIA Telescope Network

Author: López Casado, Carmen,Pérez-del-Pulgar-Mancebo, Carlos Jesús,Muñoz-Martínez, Víctor Fernando,Castro-Tirado, Alberto J.
Year: 2017
Source: https://riuma.uma.es/xmlui/bitstream/10630/14756/1/glsch-observation-scheduler.pdf
GlSch: Obse a ion Schedule o he GLORIA Telescope
Ne wo k
Ca men L´opez-Casadoa, Ca los J. P´e ez del Pulga a, V´ıc o F. Mu˜noza, and Albe o
Cas o-Ti adob
aDepa amen o de Ingenie a de Sis emas y Au om ica, Uni e sidad de Mlaga. Campus de
Excelencia In e nacional Andaluca Tech.
bDepa amen o de Fsica Es ela , Ins i u o de As o sica de Andaluca,
ABSTRACT
This pape p oposes he design and de elopmen o a schedule o he GLORIA elescope ne wo k. This ne wo k,
which main objec i e is o make as onomy close o ci izens in gene al, is o med by 18 elescopes sp ead o e
ou con inen s and bo h hemisphe es. Pa o he managemen o his ne wo k is made by he ne wo k schedule .
I ecei es he obse a ion eques s made by he GLORIA use s and hen sends i o he mos sui able elescope.
A key module o he ne wo k schedule is he elescope decision algo i hm ha makes possible o choose he bes
elescope, and hus a oiding o e ing an obse a ion o a elescope ha canno execu e i . This pape shows wo
di e en elescope decision algo i hms: he i s one is only based on wea he o ecas , meanwhile he second one
uses uzzy logic and in o ma ion om each ne wo k elescope. Bo h algo i hms we e deployed in he GLORIA
ne wo k. The achie ed esul s coupled wi h a compa a i e o hei pe o mance is shown. Mo eo e , he ne wo k
schedule a chi ec u e, based on a hyb id dis ibu ed-cen alized schema, is de ailed.
Keywo ds: As onomy, elescope ne wo k, schedule algo i hm, uzzy logic, so wa e
1. INTRODUCTION
O e he las decades, he e has been an inc ease in elescope ne wo k implemen a ions. The e is a huge
in e es in making as onomy close o s uden s and ci izens in gene al [1,2]; p o iding easy- o-use web ools o
awake hei in e es . On he o he hand, elescope ne wo ks ocused on p o essional as onome s ha e also been
de eloped [3,4].
Main pa o hese ne wo ks is based on a coo dina o o schedule ha manages all he eques s. These
eques s will include in o ma ion abou he a ge o obse e and also how o obse e i , i.e. he cons ain s.
Depending on his da a, he schedule will p opose he obse a ion o a speci ic elescope and a e wa ds, i will
execu e he obse a ion i sel . The design and implemen a ion o he schedule in he ne wo k can be dis ibu ed
o ully cen alized. The dis ibu ed a chi ec u e [5] usually includes a cen al node ha decide he elescope
and local nodes ha schedule all he ecei ed obse a ions. On he o he hand, cen alized sys ems [6] schedule
all he elescope obse a ions in he cen al node. The elescope only ac s as a sequence , ecei ing ins uc ions
and gene a ing da a.
In he las e m o 2014, he GLORIA ne wo k [1] was launched wi h he aim o c ea e a wo ld-wide elescope
ne wo k whe e use s could esea ch in as onomy using obo ic elescopes, and/o analyzing da a which o he
use s ha e acqui ed. I in eg a es elescopes ha ha e been wo king in a ious scien i ic ields and dissemina ion
issues, c ea ing a he e ogeneous elescope ne wo k. As a pa o his ne wo k, a schedule has been designed
and i is being implemen ed and deployed. This schedule manages all he obse a ion eques s and decide he
elescope ha will be o e ed he obse a ion. This decision is made h ough a elescope decision algo i hm.
This pape de ails he GLORIA ne wo k in sec ion 2, as well as he a chi ec u e o he schedule in sec ion
3. The di e en elescope decision algo i hms a e explained in sec ion 4. Nex , he s a is ic esul s o e he
ne wo k usage a e shown in sec ion 5. Finally, conclusions a e in sec ion 6.
2. THE GLORIA TELESCOPE NETWORK
The GLORIA ne wo k in eg a es 18 elescopes wo ldwide (Figu e 1) ha had been wo king in a ious scien i ic
ields and dissemina ion issues, c ea ing a he e ogeneous ne wo k whe e elescopes ime is sha ed be ween owne s
and he GLORIA ne wo k.
Figu e 1. Geog aphic loca ion o he GLORIA elescopes.
The unde lying idea in GLORIA is ha he mo e eyes we would pu on he sky he mo e and g ea e he
scien i ic disco e ies ha will be achie ed. Taking ad an age o collec i e in elligence [7] he huge amoun o
as onomical da a can be analyzed by bo h p o essional and ama eu as onome s. Thus, in o de o y o
imp o e he way o doing as onomy esea ch, use s will be guided h ough he di e en asks he esea ch
equi es. Each kind o esea ch has been called expe imen . The e a e wo kinds o expe imen s: hose ha
equi e a elescope, which ha e been called on-line expe imen s and o -line expe imen s which wo k on da a
p oduced by he GLORIA ne wo k. The a ailable o -line expe imen s a e: Pe sonal Space, whe e you choose an
impo an e en in you li e and see he sky ha was di ec ly o e head a ha exac momen ; o Sola Ac i i y,
whe e you can con ibu e o he measu emen o sola ac i i y; and Va iable S a s ha allows he analysis o
he ligh cu es o he s a s.
On he o he side, on-line expe imen s a e also di ided in o wo ca ego ies: eleope a ion and ba ch. The
i s one is used o di ec ly con ol he elescope a a speci ic da e and ime, which is chosen by he use . A ha
da e, use will be able o con ol he elescope moun o poin a a speci ic a ge , selec he il e , ocus and se
he exposu e ime o ake an image ha can be inally downloaded. On he o he hand, in ba ch mode, use s do
no con ol he elescope; ins ead, a eques o obse e a a ge is submi ed o he ne wo k and he schedule
will manage he obse a ions.
3. GLSCH: OBSERVATION SCHEDULER
Mos o he elescope ne wo ks ha a e cu en ly in use o ha a e being de eloped ha e a common cha ac e is ic:
hei elescopes a e only used by he ne wo k and hey ha e he same con ol sys em. Howe e , one o he key
ac o s o he GLORIA ne wo k is ha he elescope usage in he ne wo k is sha ed wi h he elescope owne ; he
idea is o o e , a leas , he elescope idle ime o he GLORIA communi y. This p i a e and public coexis ence
make manda o y he use o he elescope own con ol sys em and he local schedule p o ided wi h i . This ac
makes he GLORIA ne wo k o be he e ogeneous no only because i s elescopes ha e di e en ea u es, bu also
because he e a e di e en con ol sys ems o manage hem (RTS2 [8], ACP [9] ...). Mo eo e , o combine he
exis ence o a ne wo k schedule o manage he en i e use eques s, and also keep he elescope au onomy, he
elescope local schedule has o belong o he GLORIA schedule a chi ec u e.
The GLORIA schedule manages all he use eques s, aking in o accoun he cons ain s o be sa is ied and
deciding o which elescope he obse a ion will be o e ed. Then, he own elescope schedule will include he
obse a ion wi hin he nigh plan which has o be execu ed. The igu e 2shows he schedule hyb id dis ibu ed-
cen alised a chi ec u e. I is based on 3 laye s: he uppe one o med by a unique node he Cen al Node. The
middle laye whe e he e is a Local Node o each elescope in he ne wo k and inally he lowe laye whe e he
elescopes hemsel es a e loca ed.
Figu e 2. GlSch a chi ec u e.
3.1 Obse a ion managemen p ocess
The i s s ep o he obse a ion managemen p ocess is made by he Cen al Node, once he use has submi ed
he obse a ion i has o assu e ha he eques is complian wi h he eques schema. Then a cons ain
analysis is made. The cons ain s a e di ided in o ime cons ain s, ha d cons ain s and isibili y cons ain s.
The analysis o he i s wo ones is made by he Cen al Node. Time cons ain s de ine a ime window whe e he
obse a ion has o be pe o med. I his window is wi hin he ma gins he schedule is p ocessing, he eques
analysis will con inue; i no , i will be schedule la e . Nex cons ain s o be checked a e he ha d ones; hese
a e cons an o can be conside ed as cons an om he poin o iew o he elescope/obse a o y. The isibili y
cons ain s depend on he exac ime he a ge will be obse ed, e.g. a ge al i ude. The Cen al Node will
communica e wi h he local nodes o p o ide hem all he in o ma ion needed o make a isibili y s udy and
decide i he a ge can be obse ed o no . Finally, he Cen al Node, h ough he decision algo i hm, will
choose he elescope ha will be o e ed he obse a ion.
The Local Node, associa ed o he chosen elescope, no i ies he obse a ion o he elescope local schedule .
The ways o submi ing a new obse a ion o he hi d-pa y schedule s includes om c ea ing and uploading a
new ile (ACP) o use an speci ic in e ace o c ea e and hen schedule a a ge (RTS2, some cus om sys ems).
Then he elescope schedule manages he obse a ion including i in o i s nigh plan. The esul o his p ocess
can be posi i e when he obse a ion images ha e been aken; o nega i e, when he obse a ion has no been
execu ed. I local elescope schedule comple es he obse a ion, he local node e ches he images and a no i-
ica ion is sen o he cen al node which upda es he eques s a e. On he o he hand, when no images ha e
been aken, a ealloca ion p ocess, o choose ano he elescope, is s a ed.
4. TELESCOPE DECISION ALGORITHM
The di e ence in he a chi ec u e and usage o he ne wo k makes he schedule main goal o be also di e en .
While minimizing he ime pe obse a ion is he main objec i e in mos o he elescope ne wo k, he p incipal
GLORIA objec i e is o maximize he numbe o comple ed obse a ions. As each GLORIA elescope has i s
own con ol sys em wi h i s own local schedule , he obse a ion ime canno be con olled. Each GLORIA
eques will be included in he elescope nigh plan and GLORIA does no ha e any con ol abou he execu ion
ime slo . To a oid o e ing an obse a ion o a elescope ha canno execu e i and hen making he o e all
p ocess longe , he elescope decision algo i hm is c ucial.
Nex , he wo elescope decision algo i hm ha has been designed and implemen ed a e explained.
(a) Accep ance Ra e (b) Realloca ion s eps
Figu e 3. Algo i hm pe o mance compa a i e
4.1 Wea he o ecas algo i hm
This elescope decision algo i hm is based on wea he o ecas . This in o ma ion has been aken om 7 ime [10],
a ee p ojec ha p o ides speci ic in o ma ion o as onomy. The elescope decision algo i hm ecei es he
wea he o ecas o e e y elescope ha passed he cons ain s. Fo each o hem, a sco e is compu ed aking
in o accoun he wea he condi ion in i s nigh hou s, he pa ame e s conside ed a e he wea he o ecas (cloud
co e and p ecipi a ion ype) and he seeing.
4.2 Fuzzy logic algo i hm
This elescope decision algo i hm uses a uzzy logic model o sco e each a ailable elescope. The uzzy model uses
di e en kind o inpu pa ame e s: as onomical wea he in o ma ion, a ge quali y obse a ion and elescope
ne wo k eedback.
The as onomical wea he pa ame e s in o ms abou he wea he condi ion a he obse a o y loca ion:
wea he o ecas and seeing as hey we e used in he wea he o ecas algo i hm. The pa ame e used o measu e
he a ge quali y a he obse a o y loca ion is he Ta ge T ansi Al i ude. Finally, he e a e wo elescope
ne wo k eedback pa ame e s o include in o ma ion abou he p e ious pe o mance o he elescopes. The
Use Sco e, once an obse a ion is comple ed, GLORIA use s sco e he image aken by he elescope. And he
Telescope Accep ance Ra e, he pe cen age o execu ed obse a ions o e he o al o e ed ones.
5. RESULTS
The schedule ha has been desc ibed coupled wi h he de ailed decision algo i hms we e deployed and es ed
in o he GLORIA ne wo k. The expe imen consis ed o analyzing he in o ma ion p oduced by he ne wo k
du ing wo pe iods. In he i s one, he wea he o ecas algo i hm was used, meanwhile in he second pe iod
he uzzy logic one was ac i e.
Figu e 3a e eals ha he accep ance a e o bo h ype o algo i hm a e qui e simila : 37,96% when he
wea he o ecas algo i hm was in used and 43,58% when he uzzy logic one was ac i e. Howe e , a deepe
analysis on comple ed obse a ion (Figu e 3b), speci ically in he numbe o ealloca ion s eps, shows ha when
uzzy logic algo i hm is used he 77% o he comple ed obse a ion we e made by he i s elescope he algo i hm
chose e sus he 56% associa ed o he wea he based algo i hm. Tha means ha he o e all obse a ion ime
we e educe when using he uzzy logic algo i hm, exac ly i was mo ed om 41 hou s o less ha 28.
6. CONCLUSIONS
This pape has desc ibed he hyb id dis ibu ed-cen alised a chi ec u e o he GLORIA schedule , as well as
he wo di e en elescope decision algo i hms ha has been implemen ed and es ed in he GLORIA ne wo k.
A compa a i e be ween bo h algo i hms has been made and shows ha he uzzy logic algo i hm educes he
ealloca ion p ocess and hus he o e all ime o comple e he obse a ion.
ACKNOWLEDGMENTS
This wo k has been pa ially suppo ed by he GLORIA p ojec . GLObal Robo ic elescopes In elligen A ay
o e-Science (GLORIA) is a p ojec unded by he Eu opean Union Se en h F amewo k P og amme (FP7/2007-
2012) unde g an ag eemen numbe 283783.
REFERENCES
[1] Cas o-Ti ado, A. J., S´anchez Mo eno, F. M., P´e ez del Pulga , C., Az´oca , D., Beskin, G., Cabello,
J., Cedazo, R., Cues a, L., Cunni e, R., Gonz´alez, E., Gonz´alez-Rod ´ıguez, A., Go osabel, J., Hanlon,
L., Hudec, R., Jakubek, M., Janecek, P., Jel´ınek, M., La a-Gil, O., Lin o , C., L´opez-Casado, M. C.,
Malaspina, M., Mankiewicz, L., Mau ei a, E.; Maza, J., Mu˜noz-Ma ´ınez, V. F., Nicas o, L., OBoyle, E.,
Palazzi, E., P´a a, P., Pio, M. A., P ouza, M., Se ena, F., Se a-Rica , M., Simpson, R., Sp imon , P.,
S obl, J., Topinka, M., Vi ek, S., and Za necki, A. F., “The GLObal Robo ic elescopes In elligen A ay
o E-Science (GLORIA),” Re is a Mexicana de As onom´ıa y As o ´ısica 45, 104–109 (2014).
[2] G esham, K. C., Palma, C., Polsg o e, D. E., Chun, F. K., Della-Rose, D. J., and Tippe s, R. D., “Educa ion
and ou each using he alcon elescope ne wo k,” Ac a As onau ica 129, 130–134 (2016).
[3] Bakos, G. and Gaspa , “Finding and Cha ac e izing a La ge and Di e se Popula ion o T ansi ing Exoplane s
wi h HATSou h in Suppo o NASA Space Missions,” NASA P oposal #16-XRP16-70 (2016).
[4] Bigongia i, C. and Conso ium, C., “The Che enko Telescope A ay,” Nuclea and Pa icle Physics P o-
ceedings 279, 174–181 (2016).
[5] Race o, E., Oca˜na, F., and Ponz, D., “Towa ds an au onomous elescope sys em: he Tes -Bed Telescope
p ojec ,” Highligh s o Spanish As ophysics , 828–833 (2015).
[6] Saunde s, E. S., Lampoudi, S., Lis e , T. A., No bu y, M., and Walke , Z., “No el scheduling app oaches
in he e a o mul i- elescope ne wo ks,” in [SPIE As onomical Telescopes+ Ins umen a ion], 91490E (8
2014).
[7] Wang, F., Deng, H., Guo, L., and Ji, K., “A Su ey on Scien i ic-Wo k low Techniques o E-science in
As onomy,” in [2010 In e na ional Fo um on In o ma ion Technology and Applica ions], 417–420, IEEE
(7 2010).
[8] Kub´anek, P., “S a us, upg ades, and ad ances o RTS2: he open sou ce as onomical obse a o y man-
age ,” in [SPIE As onomical Telescopes+ Ins umen a ion], 99132U, In e na ional Socie y o Op ics and
Pho onics (7 2016).
[9] Denny, R., “A Web-Remo e/Robo ic/Scheduled As onomical Da a Acquisi ion Sys em,” in [Telescopes
om A a Con e ence], 47 (2011).
[10] Ye, Q.-Z., “Fo ecas ing Cloud Co e and A mosphe ic Seeing o As onomical Obse ing: Applica ion and
E alua ion o he Global Fo ecas Sys em,” Publica ions o he As onomical Socie y o he Paci ic 123, 113
(2011).