Learning Based Resource Management in Mobile Cloud Computing

Master of Science, Graduate Program in Mechanical Engineering
University of California, Riverside, August 2016
Dr. Shaolei Ren, Chairperson

Nowadays,  smart  mobile  device  such  as  smart  phones,  tablets  are playing  more  and  more  important  roles  in  our  life.  Due  to  some resource limitations  such  as  storage  capacity,  battery  lifetime,  or processing  ability,  smart  mobile  device  are  consuming  much  more energy  than  before. According  to  recent  researches,  Mobile  Cloud Computing (MCC) is becoming a promising approach, and the core of MCC  is  computational  offloading.  In  this  thesis,  we  start  from choosing different key features to make better prediction by Support Vector Machine (SVM) method to identify if a task is suitable for offloading,  then  we  simulate  two  cases  to  calculate  energy consumption,  time  cost  and  energy  saving  rate  and  make  the appropriate  decision  based  on  the  trade-off  between  the  energy consumption and time cost with the help of SVM.