13646

12409

كارشناسي ارشد

هوش مصنوعي

اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر

۱۳۹۷

دوازده، ۷۹ص.: مصور، جدول، نمودار

محمدحسين منشئي

مهران صفاياني، مجيد نبي

1397/04/11

كتابنامه

برق و كامپيوتر

مهندسي برق و كامپيوتر

ID12409

Collision avoidance for M2M communications in LTE LTE A random access channel A reinforcement learning approach Maryam Amin Jafari maryam aminjafari@ec iut ac ir 2018 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Dr Mohammad Hossein Manshaei manshaei@cc iut ac ir Abstract In the recent years the deployment of smart devices that can autonomously connect to the internet has become a newdynamic area of research which is known as Machine to Machine M2M communications M2M network is a network ofmachin type communication devices MTD which make smart decisions based on the transferred data over the internet M2M communications is one of the key enablers of the Internet of Things IoT The most viable option of implementingM2M communications is over cellular networks such as LTE LTE A However these networks have been mainly used tosupport Human to Human H2H communications which is characterized by smaller number of connection requests andlonger connection times While in M2M networks it is possible that a huge part of MTDs try to connect simultaneously totransmit only a minimum amount of data The random access channel of LTE LTE A suffers from congestion and overload ing in the presence of the thousands of MTDs Therefore one of the key challenges is the need to enhance the operationof RACH of LTE LTE A The different alternatives are provided to overcome the congestion during the random accessprocedure The most of the solutions are based on the base station operation and therefore they need to change the cellularnetwork standards In this thesis we propose a distributed solution based on reinforcement learning which extends the uplink resources forrandom access procedure along the time to overcome the congestion on RACH Different from the centralized methods inour distributed methods the MTDs autonomously and independently learn their optimal actions to minimize the collisionsover RACH We compare the performance of the proposed method with the baseline approache in the LTE standards interms of energy consumption average delay success probabilty and collision probability Simmulation results show thatthis method significantly lowers the collision probability and reduces the energy consumption of the MTDs in their accessrequest procedure Key Words M2M communications LTE LTE A networks Random access channel RACH Ma chine learning

محمدحسين منشئي

مهران صفاياني، مجيد نبي