|
Energy-Aware Opportunistic Charging and Energy
0 Z& v: R! @- y" E% Q0 ?Distribution for Sustainable Vehicular Edge and Fog
) g8 Q# l# T# `1 j4 [9 K7 F/ @8 GNetworks ! V8 k3 n1 ~. a
2 ]5 l5 U6 }- w4 Q
7 u4 H6 o% v& N# t: T2 P, M
The fast-growing popularity of electric vehicles $ \$ C* c: x( N; d- d
(EVs) poses complex challenges for the existing power grid
6 `/ `0 m7 [$ b' }- y0 @infrastructure to meet the high demands at peak charging hours. 9 i- U6 k8 Z! G( i3 k( s
Discovering and transferring energy amongst EVs in mobile
4 e% c8 B+ R0 D2 G4 p( k9 gvehicular edges and fogs is expected to be an effective solution for % p' [2 B& q x) V
bringing energy closer to where the demand is and improving the $ A: J! O4 W" }# B' v+ ~
scalability and flexibility compared to traditional charging
$ C1 w# e# O! Jsolutions. In this paper, we propose a fully-distributed energy
, W6 [! g+ s/ J; R/ n3 zaware opportunistic charging approach which enables distributed 1 T; U- R" K$ S8 g+ ]
multi-layer adaptive edge cloud platform for sustainable mobile . l4 ] _) I( l7 T2 V# H
autonomous vehicular edges which host dynamic on-demand 7 `5 T. f/ S ~
virtual edge containers of on-demand services. We introduce a - w+ N- O% y, K# S* s0 i* T( {
novel Reinforcement Learning (Q-learning) based SmartCharge
+ D; Z$ D4 Z' O) k+ e: U. ^algorithm formulated as a finite Markov Decision Process. We : Z7 e3 ~ u0 k
define multiple edge energy states, transitions and possible actions
5 v% _/ l, \+ Z! ^1 ^8 d+ yof edge nodes in dynamic complex network environments which - _) G( F, E$ ?. `; W1 H2 l
are adaptively resolved by multilayer real-time multidimensional
% X" w5 @6 d3 S$ o3 spredictive analytics. This allows SmartCharge edge nodes to more
# Q* ?( H/ T$ q) t) G2 i" Haccurately capture, predict and adapt to dynamic spatial-temporal |* g$ ], q& K: V$ b
energy supply and demand as well as mobility patterns when
7 f2 y, Y! g+ u" k/ b! cenergy peaks are expected. More specifically, SmartCharge edge % c) }8 \) U' N/ y5 o8 R6 `- O, F
nodes are able to autonomously and collaboratively understand
( D! H: n- x j# `9 }( M9 x: A# wwhen (how soon) and where the geo-temporal peaks are expected ! f7 r9 ?! e, M
to happen, thus enable better local prediction and more accurate
$ T2 c. {, c: T- H% }global distribution of energy resources. We provide multi-criteria 0 B6 }8 s( V7 B2 m6 F. `8 Y; ?
evaluation of SmartCharge against competitive protocols over 0 {/ B# W+ j7 j! W- {7 @$ b
real-world San Francisco Cab mobility traces and in the presence ; d! Y0 c1 ?9 }# k0 E
of real-world users’ energy interest traces driven by Foursquare
9 N! I& A7 V0 `; YSan Francisco dataset. We show that SmartCharge successfully
! U% e) f7 W: u2 e6 c0 l6 upredicts and mitigates congestion in peak charging hours, reduces 8 N, X0 H/ H7 q' Q, D- _
the waiting time between vehicles sending energy demand requests " a0 j. G& t7 j- K& W
and being successfully charged as well as significantly reduces the
9 C* T& Z0 A! y) Ftotal number of vehicles in need of energy. 5 u6 X& X: t. E% T& S5 H8 |( R
& Q8 q" a3 N2 o1 r) o8 x2 m
6 c; B) J) \0 ~! H- j7 g/ F* `, {
" X' N- _( k( X; X2 x
| 
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
