20.500.12556/RUL-116035 Analiza možnosti kontekstno-odvisnega prilagajanja video dekodiranja z namenom učinkovite rabe virov mobilne naprave Analysis of possibilities for efficient use of mobile resources based on context-dependent adjustment of video decoding quality Baterije se v primerjavi s preostalo strojno opremo razvijajo mnogo počasneje, zato je pomembna njihova učinkovita raba. Predlagamo sistem, ki izkoristi sposobnosti pametnih telefonov za zaznavanje in glede na uporabnikov kontekst prilagodi kakovost videa, ne da bi pri tem zmanjšal zadovoljstvo uporabnika. Z laboratorijskimi meritvami dokažemo, da je posledica boljše kakovosti videa večja poraba električne energije. Z nenadzorovanim eksperimentom smo raziskali vpliv med uporabnikovo fizično dejavnostjo in njegovo percepcijo videa. V ta namen smo razvili aplikacijo, ki zajema podatke o uporabnikovi fizični aktivnosti in njegovi izbiri ločljivosti videa. Naknadna nadzorovana študija je dodatno potrdila ugotovitve o vplivu fizične aktivnosti na percepcijo videa. Odkrili smo, da bi predlagana rešitev prihranila najmanj štiri odstotke baterije dnevno. Izpostavili smo možnosti uporabe približnega računanja in utemeljili zakaj bi bilo potrebno zaznavanje razširiti na širši kontekst. Batteries are developing much slower compared to other computer hardware, therefore it is important to use them efficiently. We proposed a system that takes advantage of smartphones' sensing possibilities, and based on user context adapts video quality without lowering user experience. With rigorous measurements we proved that better video quality results in higher energy consumption. With an unsupervised experiment we explored the relationship between user's physical activity and their video perception. In order to carry out such an experiment, we developed an application which collects data about user's physical activity and the desired video decoding quality. Additional supervised experiment further confirmed our findings on the relation between physical activity and video perception. We found that the proposed solution would save at least four percent of the battery charge per day. We discussed the potentials of approximate computing techniques and explained the need to further expand context sensing. mobilno zaznavanje kontekstno-odvisno računanje približno računanje video dekodiranje mobile sensing context-aware computing approximate computing video decoding true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2020-05-08 12:55:01 2020-05-08 12:55:11 2024-01-08 11:47:57 0000-00-00 00:00:00 2020 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 25727 14495491 Fajfar_Tine_-_Analiza_moznosti_kontekstno-odvisnega_prilagajanja_video_dekodiranja_z_namenom_uci.pdf Fajfar_Tine_-_Analiza_moznosti_kontekstno-odvisnega_prilagajanja_video_dekodiranja_z_namenom_uci.pdf 1 F456B4D12A6AC0EC7F026AA9136FC689 807bff4970d31a14abd87d8dbbe51be16ed0220550766c0d4c430c757317e2c0 cd32f1aa-a1b8-11eb-a523-00155dcfd717 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=129431 Fakulteta za računalništvo in informatiko 0 0 0