C-DAX is funded by the European Union's Seventh

Framework Programme (FP7-ICT-2011-8) under

grant agreement n° 318708

A Publish/Subscribe Communication

Framework for Managing Electric

Vehicle Charging

Dr Yue Cao,

University of Surrey

IEEE ICCVE 2014

3-7 Nov 2014

Outline

Research Overview and Motivation

Introduction on Our Proposal

Research Analysis and Method

Results Evaluation and Observations

Conclusion

Research Overview

Majority of works on Electric Vehicle (EV) charging management mainly consider static scenario

• Schedule EVs charging while they have been parking at home/charging station (when/whether to charge)

Home based charging scenario

Smart charging scenario

A few works consider EV charging management under mobile scenario

• EVs know where they should travel for charging, while they are on-the-move (where to charge)

Research Motivation on Communication Side for

On-the-Move EV Charging

Cellular network communication

• Point-to-point communication (need to know IP addresses of end users)

• Expensive even with wider radio coverage range

• Unnecessary to have ubiquitous radio coverage for on-the-move EV charging

CS-selection on where to charge is not always required

Publish/Subscribe (P/S) communication with Road Side Unit (RSU)

• Scalability

Decouple communication in space – do not need to know IP addresses of publishers and subscribers

Decouple communication in time – be tolerant to intermittent communication in VANETs, by appropriately deploying RSUs

• Reduce bandwidth usage – cost efficiency

Introduction of On-the-Move EV Charging

EVs generate charging requests while they are on-the-move during journeys

Application scenarios for on-the-move EV charging management

• Centralized decision-making (adopted by previous works)

Decision made by the global aggregator – where to charge

Ubiquitous communication – realtime information

High privacy sensitivity – EV status information (e.g., ID, location) will be released when sending charging request to aggregator

• Fully distributed decision-making (our proposed)

Decision made by individual EVs – where to charge

Opportunistically accessed communication – historical information

Low privacy sensitivity – no need to release EV status information

Network Entities

Electric Vehicle (EV): Each EV is with a Status Of Charge (SOC). Ifthe ratio between its current energy and maximum energy isbelow the value of SOC, the EV will need to request for acharging service by a CS. Here, EV is as subscriber.

Charging Station (CS): Each CS is deployed at a certain locationfor energy transaction. Here, CS is as publisher.

Road Side Unit (RSU): The deployed RSU behaves as anintermediate entity for bridging the information flow exchangebetween EVs on the road and the grid infrastructure, throughwireless communications.

Overview of Distributed Scenario

Each CS periodically publishes its condition information (e.g.demand load, EV queuing time) towards EVs through their nearbyRSUs.

Upon passing through each RSU, EVs will record this CS condition information via RSU.

• Push mode: An EV will opportunistically receive the CS publication, only when there is a CS update published while the EV is within an RSU radio coverage

• Pull mode: RSU can cache historically published CS update so that as long as an EV passes through an RSU radio coverage, it can fetch the latest update

An EV reaches its threshold in relation to SOC, starts to select the best CS using its locally recorded information.

• EV travels towards its selected CS for charging

Push mode vs. Pull mode

RSU

Published

data

RSU RSU

Published

data

RSU

Push Mode:

Periodically published data instantly advertised

to EVs in RSU coverage

Pull Mode:

Periodically published data cached at

RSU, and EVs fetch latest cached

data by sending queries

Analysis on Probability For EV to Obtain

Information From RSU

Push Mode

Pull Mode

Given 2R<S (radio range of adjacent RSUs are not overlapping)and R=L, Pull Mode achieves a higher possibility to obtaininformation, than Push Mode.

Assumption

EV passes through all RSUs evenly deployed

on straight road

Notations

V: Constant EV Speed

R: RSU Transmission Range

L: EV Transmission Range

S: Distance Between Adjacent RSUs

F: Distance From Starting Point

N: Number of RSUs

T: CS Publication Interval

𝑷𝒑𝒖𝒔𝒉 = 𝟏 − (𝟏 −𝑭 + 𝑹

𝑽 ∗ 𝑻)｛𝟏 − [(

𝟐𝑹

𝑺)(𝟐𝑹

𝑽 ∗ 𝑻)]｝(𝑵−𝟏)

𝑷𝒑𝒖𝒍𝒍 = 𝟏 −

𝒊=𝟏

𝑵

{𝟏 − [𝒊 − 𝟏 𝑺 + 𝑭 + 𝑳

𝑽 ∗ 𝑻]}

Simulation Validation

Both Modes Benefit From Slow V

Push Mode Benefits From Large R

Push Mode Benefits From Small S

Pull Mode Benefits From Large S

Both Modes Benefit From Frequent T

Pull Mode Benefits From Large L

Calculation of Queuing Time

Queuing Time = Minimum charging time of those under charging + Aggregated value of those waiting for charging

The CS at which there is a large number of EVs parking will beestimated with a long queuing time, in particular that applyingmore charging slots contributes to a short queuing time.

Performance Evaluation

CSs and RSUs are connected via wired line

Each CS is connected to all RSUs

Parameter Value

Number of EVs 100

Number of CSs 5

Number of RSUs 7

EV Moving Speed 5-10 m/s

Radio Range of RSU 300m

Radio Range of EV 300m

CS Update Interval 100s

Status of Charge 60%

EV Initial Energy 30KWh

Maximum EV Moving Distance

161KM

CS Initial Energy 3000 KWh

CS Charging Power 62KW

Results

Observations1: Performance given frequent publication isclose to that under centralized manner.

2: Frequent publication improves informationfreshness.

3: Waiting time for charging, utilization of CSsbenefit from a good information freshness .

Ideal Case: Decision making in centralisedmanner, via realtime CS queuing time

Push/Pull Mode: Decision making indistributed manner, via historical CS queuingtime

Conclusion and Future Work

Present a Publish/Subscribe communication framework for CS-selection decision making at EV side.• Low privacy sensitivity

• Moderate charging performance given carefully set publication interval

• Cost-efficiency – no need for ubiquitous communication

In future, the remote reservation service will be integrated.• EVs report their charging reservation for selected CSs, including their

arrival time at selected CSs, and how long their charging will be

• CS-selection decision to find the CS with the minimum expected waitingtime (considering local queuing EVs and those making reservation)

• EVs charging is coordinated by considering CSs condition in a near future

Question?