IEEE EDGE 2022, Barcelona - IFTA presents Paper on July 14

Pre­sen­ta­tions on the topic "Ben­e­fits of Edge and Cloud"

Roman Karl­stet­ter from IfTA presents his paper "Query­ing Distribut­ed Sen­sor Streams in the Edge-to-Cloud Con­tin­u­um" at the re­search con­fer­ence IEEE EDGE 2022 in Barcelona. July 14, 2022, 10:45 - 12:00 a.m., UPC Ter­ras­sa Cam­pus, Room 10, EDG_SHT_014

The paper by Roman Karl­stet­ter, Robert Wid­hopf-Fenk and Martin Schulz is a joint work of IfTA and CAPS (chair of the TUM School of Com­pu­ta­tion, In­for­ma­tion and Tech­nol­o­gy) and pro­pos­es an ap­proach for re­triev­ing sen­sor data from ge­o­graph­i­cal­ly dis­trib­uted pro­cess­ing and stor­age nodes. The team works on a sen­sor stor­age frame­work In the scope of the SensE project.

More in­for­ma­tion about the project SensE

More in­for­ma­tion about the IEEE EDGE 2022 pro­gram in Barcelona

Regis­ter for 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