Distribution for Sustainable Vehicular Edge and Fog
. k6 c: k7 K0 T, u
Networks
" U3 h% I4 c8 G* G x' ~# F - ]8 w1 d$ n8 y& a! L% p9 K" g$ |# {+ P3 B7 B4 Y3 o
The fast-growing popularity of electric vehicles 1 _" j2 d6 _9 M
(EVs) poses complex challenges for the existing power grid 9 [3 q7 p$ q# h( B5 V8 j: Iinfrastructure to meet the high demands at peak charging hours. : Q7 [, c6 H5 a, o
Discovering and transferring energy amongst EVs in mobile * s$ h5 N# @$ i" {vehicular edges and fogs is expected to be an effective solution for 5 C1 o" ?0 U( M! \7 C) {& m/ ?bringing energy closer to where the demand is and improving the 9 |) L, {- U! v4 @1 Q9 ]% D" {$ L
scalability and flexibility compared to traditional charging ' X9 i5 P" c# b u
solutions. In this paper, we propose a fully-distributed energy ! b* n f# J6 _aware opportunistic charging approach which enables distributed + |" t2 T6 y2 F4 T7 v, @ |, q
multi-layer adaptive edge cloud platform for sustainable mobile " k. a- v# G& I/ t* \
autonomous vehicular edges which host dynamic on-demand - ~( \4 I: Y( F" }) ?virtual edge containers of on-demand services. We introduce a # B4 |( p8 Q( Z( ?4 `8 J$ E/ F
novel Reinforcement Learning (Q-learning) based SmartCharge ( k M2 g- [* f- I* ualgorithm formulated as a finite Markov Decision Process. We " n; l* e+ q7 g9 i/ u/ k, w
define multiple edge energy states, transitions and possible actions ' k% O/ A0 T m6 P/ \9 Hof edge nodes in dynamic complex network environments which 0 d& a# v5 o9 P
are adaptively resolved by multilayer real-time multidimensional $ U( W& j8 n2 _predictive analytics. This allows SmartCharge edge nodes to more 0 ~& y) z+ ]0 U* p6 }accurately capture, predict and adapt to dynamic spatial-temporal W! f% _6 Y3 t) x6 }6 p+ f, Y
energy supply and demand as well as mobility patterns when : z) ^4 J, I& t; a0 X/ n, W0 [
energy peaks are expected. More specifically, SmartCharge edge 3 c9 U. B& M5 ^- H N0 y. S7 m0 Bnodes are able to autonomously and collaboratively understand 3 @ _: v" n8 b0 _; \
when (how soon) and where the geo-temporal peaks are expected ) L' S8 p, o9 D7 e& Q6 Cto happen, thus enable better local prediction and more accurate " a) s, S% T, }* b; e
global distribution of energy resources. We provide multi-criteria + V9 {" ^; i% D' |
evaluation of SmartCharge against competitive protocols over , D, N T2 J6 Z# l0 Wreal-world San Francisco Cab mobility traces and in the presence ; e0 c$ r7 L* e2 j* o7 zof real-world users’ energy interest traces driven by Foursquare / W6 b, y+ t L+ p+ L
San Francisco dataset. We show that SmartCharge successfully 5 T/ z4 ^3 ~$ b% L: Y
predicts and mitigates congestion in peak charging hours, reduces 8 g$ s2 f' I+ S( v+ c) n- P* ?2 E
the waiting time between vehicles sending energy demand requests / E: E& [# M' t! \. U; T5 G+ A
and being successfully charged as well as significantly reduces the 8 |) c! `. r! Q0 P- j) utotal number of vehicles in need of energy. 3 g8 `. {. _: H! Q r - O# T7 B4 F; {. y 0 c+ ~% Z2 B1 l, ?4 [ q$ P( w3 o
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
