|
Energy-Aware Opportunistic Charging and Energy
% p# U6 u7 o7 nDistribution for Sustainable Vehicular Edge and Fog
. ^/ k/ x; P' k [, UNetworks % h c' V/ J/ K$ o. J2 C
* D/ r# h Y( d$ ^% k- q0 Y- O7 T+ p+ k4 N
The fast-growing popularity of electric vehicles
. M/ I* Z7 ~: [* k/ f(EVs) poses complex challenges for the existing power grid
0 A* r5 }1 n5 q0 C5 K( Y1 minfrastructure to meet the high demands at peak charging hours. + g0 T& k5 z: @/ ?" B
Discovering and transferring energy amongst EVs in mobile
2 y1 y- w# R c- G5 qvehicular edges and fogs is expected to be an effective solution for 4 y6 N$ A" O! O& F; }1 Y) q
bringing energy closer to where the demand is and improving the ; E' P/ a9 E( F, B9 `6 W3 H/ V
scalability and flexibility compared to traditional charging 1 @( v1 d v) ?
solutions. In this paper, we propose a fully-distributed energy
* `( `8 G4 {& Z' |: Uaware opportunistic charging approach which enables distributed
. c( X5 M7 r: c4 kmulti-layer adaptive edge cloud platform for sustainable mobile
! w( a; `/ b: H0 Pautonomous vehicular edges which host dynamic on-demand
6 @; o6 h! t7 m/ [ E! Jvirtual edge containers of on-demand services. We introduce a 0 j/ U! e7 u7 z* v R% N
novel Reinforcement Learning (Q-learning) based SmartCharge
^# i5 @- B# H: talgorithm formulated as a finite Markov Decision Process. We 0 }- J$ o/ P% B9 R6 A
define multiple edge energy states, transitions and possible actions : t$ |7 G- q' T4 w. v8 {
of edge nodes in dynamic complex network environments which % }0 W7 h* W8 X2 @
are adaptively resolved by multilayer real-time multidimensional
, m& q8 w& b. j$ J% Tpredictive analytics. This allows SmartCharge edge nodes to more & T, G+ @8 s5 N% ?0 {
accurately capture, predict and adapt to dynamic spatial-temporal * g3 D+ L- B! y) ] C% z
energy supply and demand as well as mobility patterns when 1 y X4 z; L0 e6 V6 ?2 K
energy peaks are expected. More specifically, SmartCharge edge + {7 ]5 l+ ~/ x8 t
nodes are able to autonomously and collaboratively understand
3 ~3 f, R" p- |3 f7 Swhen (how soon) and where the geo-temporal peaks are expected
8 Q x5 H% T! w# s: ^to happen, thus enable better local prediction and more accurate + i5 U" H4 G# B* c P4 R$ y3 O# A& H
global distribution of energy resources. We provide multi-criteria
" |) u0 ]# N0 O% g6 H9 `2 b5 oevaluation of SmartCharge against competitive protocols over
4 n, \: s- V t8 M2 S( b; Ureal-world San Francisco Cab mobility traces and in the presence
: p1 V) J* s7 f! Lof real-world users’ energy interest traces driven by Foursquare
( p, y# g8 E, pSan Francisco dataset. We show that SmartCharge successfully : n% k* |% e( ^- C0 s& t+ l
predicts and mitigates congestion in peak charging hours, reduces
* M7 {7 c( _1 ^; Hthe waiting time between vehicles sending energy demand requests
% G% P8 U4 f7 W- M% h; D% N& Eand being successfully charged as well as significantly reduces the 2 i+ m# ~1 F4 b8 Q4 O2 u
total number of vehicles in need of energy.
& P% ] i9 ^! z- X6 T( o: p! z1 J! n& r* }
W% a- [* x1 M/ z- w0 ?
9 T [( J- N: f* f- c+ y
; {4 ]) e. a( L/ 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
