A Mobility-aware Cross-edge Computation Offloading Framework for - PowerPoint PPT Presentation

A Mobility-aware Cross-edge Computation Offloading Framework for Partitionable Applications Hailiang Zhao 12 Shuiguang Deng 1 Cheng Zhang 1 Wei Du 2 Qiang He 3 Jianwei Yin 1 1 Zhejiang University, Hangzhou, China 2 Wuhan University of Technology, Wuhan, China 3 Swinburne University of Technology, Melbourne, Australia July 10, 2019 Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 1 / 16

Outline Introduction 1 A Brief Introudction to Mobile Edge Computing (MEC) What is the Problem? Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 2 / 16

Outline Introduction 1 A Brief Introudction to Mobile Edge Computing (MEC) What is the Problem? Cross-edge Computation Offloading 2 System Model Problem Formulation Proposed Framework and Algorithms Experimental Evaluation Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 2 / 16

Introduction Outline Introduction 1 A Brief Introudction to Mobile Edge Computing (MEC) What is the Problem? Cross-edge Computation Offloading 2 System Model Problem Formulation Proposed Framework and Algorithms Experimental Evaluation Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 3 / 16

Introduction A Brief Introudction to Mobile Edge Computing (MEC) What is Mobile Edge Computing? Mobile Edge Computing Mobile Edge Computing (MEC) is a new computation paradigm: depolyed at the network edge 1 use widespread wireless access network (Small-cell Base Station) 2 provide service and computing resource 3 Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 4 / 16

Introduction A Brief Introudction to Mobile Edge Computing (MEC) What’s it properties? Edge site An edge site is a micro data center with applications depolyed, attached to a small-cell base station (SBS). 1 Heterogeneous edge sites 2 User mobility 3 Edge site selection and user profile handover 4 Overlapped signal coverage of SBSs (Corss-edge Collaboration!) 5 Partitionable applications (data stream) 6 Insufficient battery energy of mobile devices 7 ... ... Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 5 / 16

Introduction What is the Problem? Motivation Scenario time slot #2 User #3 time slot #1 partitioning & o 225 270 315 partitioning & o time slot #1 time slot #5 time slot #4 User #1 time slot #4 User #2 time slot #1 time slot #3 time slot #3 time slot #2 0 180 o time slot #3 time slot #6 time slot #2 o time slot #6 time slot #6 time slot #5 90 135 time slot #5 45 time slot #4 Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 6 / 16

Introduction What is the Problem? Problem Definition For partitionable applications, how to make the offloading strategy with the minimum overall cost achieved? Composition of overall cost 1 execution delay 2 task dropping penalty 3 collaboration cost Energy Harvesting (EH) technology is adopted. Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 7 / 16

Cross-edge Computation Offloading Outline Introduction 1 A Brief Introudction to Mobile Edge Computing (MEC) What is the Problem? Cross-edge Computation Offloading 2 System Model Problem Formulation Proposed Framework and Algorithms Experimental Evaluation Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 8 / 16

Cross-edge Computation Offloading System Model System Model 1 Local execution latency evaluation i � η l execution latency: τ lc i /f i 1 energy consumption: ǫ l i � κ i · η l i f 2 2 i 2 Offloading latency evaluation µ r 1 transmission delay: τ tx i,j ( t ) � j ∈M i ( t ) I i,j ( t ) · i 1 � R i,j ( t ) η r execution delay: τ rc i,j ( t ) � i 2 f j · � j ∈M i ( t ) I i,j ( t ) collaboration cost: ϕ · � j ∈M i ( t ) I i,j ( t ) 3 constraint: 4 � τ tx i,j ( t ) + τ rc � + τ lc τ d ≥ max j ∈M i ( t ) i,j ( t ) i + ϕ · � j ∈M i ( t ) I i,j ( t ) 3 Battery energy level evaluation envolution function: 1 ψ i ( t + 1) = ψ i ( t ) − � j ∈M i ( t ) ǫ tx i,j ( t ) · I i,j ( t ) − ǫ l i + α i ( t ) constraint: ǫ l i + � j ∈M i ( t ) ǫ tx i,j ( t ) I i,j ( t ) ≤ ψ i ( t ) 2 Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 9 / 16

Cross-edge Computation Offloading Problem Formulation Problem Formulation A non-convex optimization problem T − 1 � � � 1 � P 1 : min lim C ( I i ( t )) , E T T →∞ ∀ i, I i ( t ) ,α i ( t ) t =0 i ∈N with several constraints. τ tx i,j ( t ) + τ rc � � � C ( I i ( t )) max i,j ( t ) j ∈M i ( t ): I i,j ′ ( t )=1 � τ lc i + ϕ · I i,j ( t ) + ̺ i · D i ( t ) + j ∈M i ( t ) Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 10 / 16

Cross-edge Computation Offloading Proposed Framework and Algorithms Proposed Framework Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 11 / 16

Cross-edge Computation Offloading Proposed Framework and Algorithms Proposed Algorithms The CCO algorithm 1 Lyapunov optimization (drift-plus-penalty) ∀ i, I i ( t ) ,α i ( t ) ∆ up P 2 : min V ( Θ ( t )) , with several constraints. N M ∆ up � ψ ′ � α i ( t ) − ǫ l ǫ tx � � i − � V ( Θ ( t )) i ( t ) i,j ( t ) I i,j ( t ) i =1 j =1 N � C ( I i ( t )) + C + V i =1 Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 12 / 16

Cross-edge Computation Offloading Proposed Framework and Algorithms Proposed Algorithms The CCO algorithm Algorithm 1 Cross-edge Computation Offloading (CCO) 1: At the beginning of the t th time slot, obtain i.i.d. random events A ( t ) , E h ( t ) � [ E h 1 ( t ) , ..., E h N ( t )] and channel state information. 2: ∀ i ∈ N , decide I ⋆ i ( t ) , α ⋆ i ( t ) by solving the deterministic problem P 2 . 3: ∀ i ∈ N , update the battery energy level ψ i ( t ) . 4: t ← t + 1 . 1 optimal energy harvesting: α ⋆ i ( t ) 2 optimal edge site selection: I ⋆ i ( t ) Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 13 / 16

Cross-edge Computation Offloading Experimental Evaluation Benchmark Policies Random Selection (RS) 1 Greedy Selection on Communication (GSC1) 2 Greedy Selection on Computation (GSC2) 3 Parameter settings Parameter Value Parameter Value τ d 2 ms ̺ i 2 ms ϕ 0 . 02 ms ρ i 0 . 6 µ l µ r 100 bits 3000 bits i i f i 1 . 5 GHz f j 32 GHz ψ safe 10 − 28 κ i 40 mJ i N max 5 ω 1 . 5 / � i ∈N j ( t ) I i,j ( t ) GHz j 10 − 13 W p tx ̟ 0 1 W i 4 . 8 × 10 − 4 J 10 − 4 E max g 0 i,h Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 14 / 16

Cross-edge Computation Offloading Experimental Evaluation Optimality and stability Overall costs in different time slots 0.16 0.15 Average cost of all mobile devices 0.14 CCO algorithm RS GSC1 0.13 GSC2 0.12 0.11 0 25 50 75 100 125 150 175 200 time slot Figure: Average cost of mobile devices. Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 15 / 16

Cross-edge Computation Offloading Experimental Evaluation Optimality and stability Average battery levels in different time slots CCO algorithm 0.0325 RS GSC1 GSC2 0.0300 0.0275 Battery energy levels 0.0250 0.0225 0.0200 0.0175 0.0150 0 25 50 75 100 125 150 175 200 time slot Figure: Average battery energy level of mobile devices. Hailiang Zhao (Zhejiang University) Cross-edge Computation Offloading July 10, 2019 16 / 16