IEEE EDGE 2022, Bar­ce­lo­na - IFTA prä­sen­tiert Paper am 14. Juli

Roman Karls­tet­ter von IFTA prä­sen­tiert sein Paper "Qu­e­ry­ing Dis­tri­bu­ted Sen­sor Stre­ams in the Edge-to-Cloud Con­ti­nu­um" auf der For­schungs­kon­fe­renz IEEE EDGE 2022 in Bar­ce­lo­na. 14. Juli, 2022, 10:45 - 12:00 Uhr, UPC Ter­ras­sa Cam­pus, Raum 10, EDG_SHT_014

Das Paper von Roman Karls­tet­ter, Ro­bert Wid­hopf-Fenk und Mar­tin Schulz ist eine ge­mein­sa­me Ar­beit der IFTA und des CAPS Lehr­stuhls der TUM School of Com­pu­ta­ti­on, In­for­ma­ti­on and Tech­no­lo­gy und schlägt einen An­satz zum Abruf von Sens­or­da­ten aus geo­gra­phisch ver­teil­ten Ver­ar­bei­tungs- und Spei­cher­kno­ten vor. Das Team ar­bei­tet ge­mein­sam an einem Sen­sor­spei­cher-Fra­me­work im Rah­men des so­ge­nann­ten SensE-Pro­jek­tes.

Mehr In­for­ma­tio­nen zum Pro­jekt SensE

Mehr In­for­ma­tio­nen zum Pro­gramm der IEEE EDGE 2022 in Bar­ce­lo­na

Zur Re­gis­trie­rung  IEEE EDGE 2022

Abstract of the paper:

Sen­sor data is of cru­ci­al im­port­an­ce in many IoT sce­na­ri­os. It is used for on­li­ne mo­ni­to­ring as well as long term data ana­ly­tics, enab­ling count­less use cases from da­ma­ge pre­ven­ti­on to pre­dic­ti­ve main­te­nance. Mul­ti­va­ria­te sen­sor time se­ries data is ac­qui­red and in­iti­al­ly sto­red close to the sen­sor, at the edge. It is also be­ne­fi­ci­al to sum­ma­ri­ze this data in
win­do­wed ag­gre­ga­ti­ons at dif­fe­rent re­so­lu­ti­ons. A sub­set of the re­sul­ting ag­gre­ga­ti­on hier­ar­chy is ty­pi­cal­ly sent to a fog or cloud in­fra­struc­ture, often via in­ter­mit­tent or low band­width connec­ti­ons. Con­se­quent­ly, dif­fe­rent views on the data exist on dif­fe­rent nodes in the edge-to-cloud con­ti­nu­um. Ho­we­ver, when que­ry­ing this data, users are in­te­res­ted in a fast re­spon­se and a com­ple­te, uni­fied view on the data, re­gard­less of which part in the in­fra­struc­ture con­ti­nu­um they send the query to and where the data is phy­si­cal­ly sto­red.

In this paper, we pre­sent a loo­se­ly cou­pled ap­proach that ena­bles fast range que­ries on a dis­tri­bu­ted and hier­ar­chi­cal sen­sor da­ta­ba­se. Our sys­tem only as­s­u­mes the pos­si­bi­li­ty of fast local range que­ries on a hier­ar­chi­cal sen­sor da­ta­ba­se. It does not re­qui­re any sha­red state bet­ween nodes and thus de­gra­des gra­ce­ful­ly in case cer­tain parts of the hier­ar­chy
are un­re­acha­ble.

We show that our sys­tem is sui­ta­ble for dri­ving in­ter­ac­ti­ve data ex­plo­ra­ti­on ses­si­ons on ter­aby­tes of data while unify­ing the dif­fe­rent views on the data. Thus, our sys­tem can im­pro­ve the data ana­ly­sis ex­pe­ri­ence in many geo-dis­tri­bu­ted sce­na­ri­os.

The work is part­ly spon­so­red by Bay­ri­sche For­schungs-Stif­tung