|
Energy-Aware Opportunistic Charging and Energy
: j8 V4 K0 G& z H7 k: k$ U+ l8 aDistribution for Sustainable Vehicular Edge and Fog
6 r' k$ g! H w# nNetworks * r) X: P% {# q/ W+ [
; x4 _5 n; j+ }2 {2 X( K
2 ^ I, J" u4 [( Y* z
The fast-growing popularity of electric vehicles * y: V; s) J) E( f
(EVs) poses complex challenges for the existing power grid 1 [5 }7 V! B5 F/ O7 g; o3 j; [9 ^0 h8 l6 M8 x
infrastructure to meet the high demands at peak charging hours.
7 H0 r8 A# F7 h6 ~3 B4 w1 PDiscovering and transferring energy amongst EVs in mobile 1 C! v5 }3 p. @- e7 _* {
vehicular edges and fogs is expected to be an effective solution for
; B3 H4 x" y; ~5 Q+ Ibringing energy closer to where the demand is and improving the ! n; L, G8 x1 b7 \& i' `
scalability and flexibility compared to traditional charging
; c9 d3 l |& X2 z" Isolutions. In this paper, we propose a fully-distributed energy
6 W/ J9 j5 z2 M: N, Raware opportunistic charging approach which enables distributed
, z& R$ ]/ Q0 U( O- l* Kmulti-layer adaptive edge cloud platform for sustainable mobile
! R# j1 K4 q1 B: S- Bautonomous vehicular edges which host dynamic on-demand
! G5 `0 Q8 v8 M0 q' W( w! B4 pvirtual edge containers of on-demand services. We introduce a " ^' Z. M B3 v5 x9 ?
novel Reinforcement Learning (Q-learning) based SmartCharge 0 v# C5 S9 G5 v/ D" [
algorithm formulated as a finite Markov Decision Process. We
/ C1 K. j8 g9 e# g2 J3 Ndefine multiple edge energy states, transitions and possible actions
1 \" @+ u( @( N; t& g$ M3 v$ R) @of edge nodes in dynamic complex network environments which 7 o# U! s$ c; I5 y7 I: }
are adaptively resolved by multilayer real-time multidimensional
" P, @' S7 ?6 R, a, Epredictive analytics. This allows SmartCharge edge nodes to more , G8 {! S# b, M
accurately capture, predict and adapt to dynamic spatial-temporal
* K$ q1 E: u& f; y( Eenergy supply and demand as well as mobility patterns when
1 n" c4 r' K- penergy peaks are expected. More specifically, SmartCharge edge
4 \/ k+ v0 B" J5 x" u& [2 `) H6 u6 rnodes are able to autonomously and collaboratively understand
& y2 e# { G9 \: Mwhen (how soon) and where the geo-temporal peaks are expected 7 i) w- K g2 ~2 o; f$ I% `
to happen, thus enable better local prediction and more accurate ( ~4 N# o& Y8 z c A! K" {1 N6 w
global distribution of energy resources. We provide multi-criteria 5 k; v5 n9 @$ Z0 A
evaluation of SmartCharge against competitive protocols over * y' p; N. l/ k, T9 Y
real-world San Francisco Cab mobility traces and in the presence
$ ?& D% b: @8 p4 Q8 Fof real-world users’ energy interest traces driven by Foursquare 8 @9 T0 e7 E( C1 B* a
San Francisco dataset. We show that SmartCharge successfully 8 q$ K0 ?" [: P. m1 n
predicts and mitigates congestion in peak charging hours, reduces ( O c- t& h1 R/ M
the waiting time between vehicles sending energy demand requests ; L: `, U* n7 p- j4 n! J
and being successfully charged as well as significantly reduces the 1 M) M: I& a% O M8 h1 y
total number of vehicles in need of energy. + s' m* \: b% B7 G$ ~
; A& R8 C5 R0 {, P3 k3 [3 h1 Y: T! ~- w" y- g: M& k
* p! F, c* N% Z. q s. h2 o* I! E0 w+ V$ A1 s* Q. H
| 
IP卡
狗仔卡
发表于 2020-11-9 15:10
转播
淘帖
分享
收藏
支持
反对
微信
提升卡
置顶卡
沉默卡
喧嚣卡
变色卡
抢沙发
显身卡
