|
Energy-Aware Opportunistic Charging and Energy
) g5 D$ y5 G3 e5 xDistribution for Sustainable Vehicular Edge and Fog 5 Z, d4 ~3 ]# ^' I& V
Networks * V5 X& s/ ?8 ~6 F* y' w; A6 l
+ T3 t$ D" k. D D
+ |4 I+ c8 ~: ]$ n0 UThe fast-growing popularity of electric vehicles
8 R0 z. u1 O, b(EVs) poses complex challenges for the existing power grid
* i( z) L9 s9 W/ `7 Winfrastructure to meet the high demands at peak charging hours. + ~/ }6 S: e7 Z* ?6 l
Discovering and transferring energy amongst EVs in mobile , q) K* M% ~! c+ d4 M
vehicular edges and fogs is expected to be an effective solution for 7 v# q, c( F ^+ y9 X# h
bringing energy closer to where the demand is and improving the
# g4 N1 u; n1 [+ I0 I* tscalability and flexibility compared to traditional charging
, T- b( A X5 f/ Ksolutions. In this paper, we propose a fully-distributed energy
5 t! r* C- p j+ N! Naware opportunistic charging approach which enables distributed
2 v4 } Q5 i6 s" F6 e+ Lmulti-layer adaptive edge cloud platform for sustainable mobile . e" Y) J5 M2 o) ]6 d; J( C; r" q
autonomous vehicular edges which host dynamic on-demand
8 k7 `; Y3 Q; z8 Dvirtual edge containers of on-demand services. We introduce a - V+ S/ B& K! T9 [7 u4 q
novel Reinforcement Learning (Q-learning) based SmartCharge : ^4 E) l1 k4 a* t+ y' M
algorithm formulated as a finite Markov Decision Process. We
& o' z( U7 t1 L; M% ^* f9 z. ^define multiple edge energy states, transitions and possible actions % t( @2 c" [1 ^7 W' @, [
of edge nodes in dynamic complex network environments which
P, S/ g+ v& r; A+ ~are adaptively resolved by multilayer real-time multidimensional
( J0 V' m( e+ ]9 }5 opredictive analytics. This allows SmartCharge edge nodes to more . \6 S% t) H+ X1 v3 D
accurately capture, predict and adapt to dynamic spatial-temporal , K9 P# y$ k% W: G8 Z0 B, c
energy supply and demand as well as mobility patterns when 9 M: @( p2 S9 ^8 Z5 Y5 E
energy peaks are expected. More specifically, SmartCharge edge
9 F6 ?# ]4 I2 g" z1 y% P* gnodes are able to autonomously and collaboratively understand % _# I$ L: C4 J2 `
when (how soon) and where the geo-temporal peaks are expected
3 t0 G3 z# g% h3 I9 Ito happen, thus enable better local prediction and more accurate 2 J6 a5 J0 [/ d4 t* c! v
global distribution of energy resources. We provide multi-criteria ; s" s, l2 C0 k6 `( B: a: w# h
evaluation of SmartCharge against competitive protocols over 9 |& Q& _: M+ I. o: L. u
real-world San Francisco Cab mobility traces and in the presence
6 O! _# s. s5 _# e, {/ iof real-world users’ energy interest traces driven by Foursquare
" Z" F! V6 ~+ vSan Francisco dataset. We show that SmartCharge successfully 7 @7 C/ }$ M. o% r- D
predicts and mitigates congestion in peak charging hours, reduces 6 [( n: z2 V1 L+ c a
the waiting time between vehicles sending energy demand requests
/ k4 C1 @1 \' U5 Sand being successfully charged as well as significantly reduces the
) c" I; a9 `" ]. Q3 s7 W2 L: gtotal number of vehicles in need of energy. : u" y1 P) R. h/ g/ o4 E
+ f8 f0 k1 R+ J" S. A% @- O
8 { u' f% ?6 B: k" N6 r: [+ ]
) k# y# q# s* N+ [
! t+ G' }2 ~4 | | 
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
