Participatory Cloud Computing

vanillaoliveInternet και Εφαρμογές Web

3 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

105 εμφανίσεις



Participatory Cloud Computing

Taariq Mullins


Jarred Martin
-

Lutando Ngqakaza

CSC4000W

UCT Computer Science Honours 2012





1

Project Description:


Cloud Computing is a rapidly growing computing paradigm
[1]

which is increasingly being
adopted by enterprise and individuals who which to utilize online services to their full potential.

Participatory Cloud Computing [2] keeps true to the idea behind cloud computing but seeks to
address its shortcomings by solving the issues that current cloud implementations. By extending
the idea and not the implementation, Participatory Cloud Computing
addresses the issues of
Cloud Computing’s growing carbon footprint, client lock in

and very lacking privacy controls.


A Participatory Cloud

extends grid computing to provide HAAS and SAAS solutions. This can
applied to provide local services as well as be
ing part of a globally connected grid provisioning
enterprise level services. This project has a local focus,

that is,

to provision local health care
professionals with tools to enable them to provide

better health care services. The focus is even
more nar
rowly defined to enable rural health care service provision. With this in mind our
network comprises of two distinct parts. A networked grid of computing devices that includes
desktop PCs, laptops, mobile phones and various smart boards and routers. The s
econd part of
the network would be remote sites that would have no connection to the grid due to lack of
internet connectivity.


Remote monitoring of healthcare patients is still a field worth
investigatin
g
[3]
;

there are many
challenges to consider when planning an infrastructure which handles the medical data of
patients
[4]
[5]
. The prop
osed project aims to apply the Participatory Cloud into the field of
monitoring health care patients, The Participatory Cloud will be used to perform operations on
sensitive medical data,
and these

operations could be machine learning algorithms
[6]

or data
mining techni
ques. The system will be used to monitor patients in rural villages where there is
limited or no data connectivity
[6]
. On arbitrary time scales the system would monitor patients and
will rank them based on those in need of critical health care.


In villages where there
is no data connectivity the system would use a form of

a

delay tolerant
network

setup
[7]
. The DTN would be comprised of wireless netwo
rks in vi
l
lages where patients
are being monitored, wireless network in major towns where there is data connectivity and
vehicle based data courier that would transport data between sites
[7]
. Where there is access to the
mobile telephone network data patient and environment data would be uploaded over the mobile
network to the regional health office for storage and analysis.



2

List of
Abbreviations:


PCC
-

Participatory Cloud Computing.

WSB
-

Wireless Smart Board (ALIX Board).

DB
-

Database / Datastore.

I²C
-

Inter
-
Integrated Circuit (Serial Bus).

SPI
-

Serial Peripheral Interface.

UART
-

Universal asynchronous receiver/transmitter.

ICT

-

Information Communications Technology.

OBWVU
-

On
-
board Wireless Vehicle Unit.

HCWU
-

Health Care Wireless Unit.

HCHS
-

Health Care Host System.

HAAS
-

Hardware as a Service

SAAS


Software as a Service


3

Problem Statement:

Poor
internet connectivity

in rural areas

has meant that networked application
deployment has

been severely stunted. This is even more the
prevalent
in developing

nations. In our specific area
of interest the issue is that patient medical data as well as environment data needs to be

collected
and
analysed
. This is not only needed for patient monitoring but also for
analysing

the
environment in relation to treatment and illness.


Rural and remote medical centres also need to know if suppliers are having shortages of drugs as
informed

decisions can be made in providing care at local sites can be better made.


Shared health data means that patients can be treated at different health facilities, since remote
clinics usually do not provide tertiary health care services
, without

having to
worry about the
patient records being transferred across to the new medical facility. Patient history is the most
crucial part of diagnosis and since South Africa still uses paper for documenting this data it
therefore becomes more difficult to distribute
this data to medical diagnosticians.


Connecting these remote locations using conventional connectivity is usually not a feasible option
as broadband capability is usually not available in these areas. Connectivity via mobile internet
technologies such as
UTMS, WiMax or Point
-
to
-
Point wireless links are expensive. Complex ICT
infrastructure also requires expert knowledge in their installation, operation and maintenance.



A successful implementation of a delay tolerant wireless data dissemination network th
at uses
motor vehicles
[7]

for the transmission of health and environmental data to the cloud is a method
proposed to address this prob
lem in the developing and non
-
developing world.


Having this information stored in a cloud VS the traditional hosting means that in addition to the
hosting benefits that Cloud Computing and traditional distributed hosting provides, one can also
provision
other services such as analysing data, investigating trends

and data warehousing.


There is also the issue that there are fewer health professionals in rural/remote areas to provide
medical assistance to patients. Having an automated
[6]

system where patients can be remo
tely
monitored and having that patient data
analysed

after which doctors are alerted should analysis
results trigger critical bounds, will allow doctors to deal with a lot more cases at once and
prioritize patients that need urgent assistance.


The project

assumes that getting more information about the patient’s health will improve the
health care being provided. We will however, not investigate the actual impact on patients as this
is beyond the scope of the project. We will however investigate the impact

on the provision of
services by medical professionals.


Current cloud computing implementations, while enabling SMEs, is adding significantly to our
carbon footprint due to its cooling requirements as well as placing noticeable strain on power
grids
[2]
.


While these services are reliable and distributed, the
y often do not provide geo
-
dependent
services. Which means African countries with low broadband penetration will have a difficult
time providing services that can give local users an experience that will be on par with users who
are located closer to their

cloud providers or have high broadband connections.




4

Procedures and Methods:


The systems architecture will be broken up into 4 parts. While each of these parts can be
seen as separate systems, each part does form part of the healthcare monitoring
system.


4.1

Healthcare sensing:

The measuring of a patient's health information would be done in the rural village of the
patient where there is no access to medical professionals and vitals measuring equipment.
The measuring of these vital parameters wou
ld have to be done by the patient using the
system.


The system would have to accommodate for the computer skills of the people using it.

The
data for this section of the project would all be “faked” at this point. Time constraints
prevent us from getting

ethical clearance for doing the actual monitoring ourselves. An
investigation is still underway to determine if we can link with other researchers in who
might be undertaking the task of monitoring and recording patient data.


4.1.1 Vital sign measurement
s

The proposed solution in measuring a patient's vital signs is to use an Android mobile
phone as a platform that has the various sensors integrated to take the measurements.

The four

most important vital parameters that could be measured are;



Body tempera
ture



Heart rate



Blood pressure



Respiratory rate


These parameters would give a general overview of a
person’s

health.


As the platform that would be used is Android the Android IOIO board would be used to
integrate the sensors with the Android mobile
phone. The IOIO board allows for sensor
peripherals to communicate over I2c, SPI and UART with the mobile phone.


4.1.2 Environmental sensing

The environment that a patient lives in is a key factor in diagnosing a particular condition
in medicine
[7]
. The system allows for measurements of environmental parameters to be
taken and stored along with a patient's medical information.


The sens
ing of environmental parameters would occur at a remote site that would be in the
village. The

remote site serves as the a central hub where patients in the village medical
data is stored so that when a vehicle equipped with the data courier unit arrives t
he
medical and environmental data can be uploaded for transfer to the regional health office.


4.1.3 Usability


Android:
Since an Android mobile phone would be used as an interface for the patient to
record vital sign parameters, the system would take
advantage of the graphical user
interface provided to allow patients who do not have a high level of computer skills to
interact with the system.


Web Front End:
The web front end would serve as a portal for the database. We will not
test for usability but

we will put a substantial amount of effort into making it as simple as
possible.


4.1.4 Data courier unit

The data courier unit would be mounted to one of the vehicles that regularly visits the site
where there are patients using this system. The data cou
rier unit works with the remote
site to transfer patient and environment data between an area that is without data
connectivity and the regional health office of the government.


4.2

Security and Privacy:

Since the nature of the data is medical we suggest
that all communications within the
system should be over encrypted network channels
[4]
. This will ensure protection against
integrity attacks and communication interception. The communications between the
wireless smart boards (ALIX boards) and the datastore/database will also be encrypte
d.
Access to the database of medical data should also be by only authorized medical
personnel and the ALIX boards,
which

requires us to set up a security policy on database
access and transactions.


Integrity checks will be performed on data communications

to make sure that all data
communicated is still intact. We will use digital certificates to enforce this
[4]
.


4.2.1

Communications Procedures and Protocols:

Communications between DB and WSB: The communication between these links will be
via untrusted links,
so the

communications should be done through a SSL socket.
MySQL
have

produced documentation on the best practices on how to do secure database
transactions and queries.
1


Communications between WSB and Medical Sensing Devices: The wireless
communication between

these two systems will also need to be secure and integrity of the
data will have to be check at communication endpoints


1

http://dev.mysql.com/doc/refman/5.0/en/security.html


4.3

Middleware and Database:

Dissemination would occur at multiple entry poin
ts. Th
is could be either via public IP

access meaning devices with internet connectivity can be connected directly possibly
giving live updates of data. This entry point can also be used to connect to remote grids
that only have periodic internet access or to extend the grid to new sites that
have
permanent internet connectivity.


This public access however requires security controls to ensure data being sent is from
authorized systems and that new systems wishing to join are clearly identified as being
uncompromised and trustworthy nodes, as s
uch nodes could insert data which could
negatively impact the analysis of data in decision making processes.


The file system will be distributed across various operating systems and device types. This
will be transparent to the connecting sensor hardware
and user accessed interfaces. The
Device types will include computers, Alix boards and Mikrotik routerboards. These
devices will be flashed with OpenW
rt
.


From the OpenWrt website:


OpenWrt is described as a Linux distribution for embedded devices.


Inste
ad of trying to create a single, static firmware, OpenWrt provides a fully writable
filesystem with package management. This frees you from the application selection and
configuration provided by the vendor and allows you to customize the device through th
e
use of packages to suit any application. For developer, OpenWrt is the framework to build
an application without having to build a complete firmware around it; for users this
means the ability for full customization, to use the device in ways never envis
ioned.


MySQL and MySQL replication will be used as the primary datastore. This will suffice
for the current scale of the system. Brewer’s CAP theorem dictates though that if we want
to scale up we would need to switch to BASE transactions
vs.

the convent
ional ACID.


Hence the choice is made that MySQL will be used as the primary database for holding
recent data. This data would then have to be warehoused in a datastore and deep web
queries would have to be facilitated in order to access and
analyse

this
data.


All the delegation of resources and access to them will be done via the resource broker.
The solution to this multidimensional problem will be tackled by knapsack problem
solving techniques, e.g. dynamic programming, meet in the middle algorithm
.


It is still unsure what sort of file system will be required to run on this sort of network.
Investigation has been made into various grid and distributed file systems. This will not be
easily extended to mobile devices such as Android or Windows Mobile. I
t is thus
suggested that
a hybrid system is developed which can accommodate devices that can run
the grid file system of choice as well as creating media access points so that mobile
devices too can join the grid.


4.4

Front End Application:

The front end application will serve as a dashboard for medical professionals to check up
reports on their patients. It will also serve as a notification
centre

for medical professionals
to monitor critical patients


5

Ethical, Professional and legal Issue
s:


There are various aspects to consider when experimenting with this type of a system. The
very nature of the data being handled wireless networks has some ethical and legal issues
to resolve.


With this in mind we have to either fake the data or seek ou
t anonymised health care data
that has been collected already and is available to be used in research.


5.1

Sensitive Data Transmission and Storage.


Since this project proposes to store sensitive medical information of individuals or testers,
ethical clea
rance will be required. We are obligated as professionals to ensure that this
data is only viewed by authorised personnel. The secure transmission of this data ensures
that the data won't be picked up by curious attackers. There will be security policies
e
nforced on the medical database as well.


5.1

Gathering of Environmental Data


Auxiliary environmental data will be collected, although the data is not of a sensitive
nature. There still needs to be clearance to conduct these experiments.

6

Related Work:


6.1

Vehicle Based Delay Tolerant Network

An experiment to determine the effectiveness of a delay tolerant network for transmission
of medical and environmental data using vehicles
[7]

was conducted by. The experiment
was conducted in two rural villages near Sorata, Bolivia. One of the villages, Quiabaya
does not have access to the fixed line telephone network or the wireless telephone
net
work;

however it is serviced by at least two daily buses
[7]
.


In the village of Quiabaya the system was setup that included a nurse's personal computer
for storing patient records and keeping track of medical supply levels as well as a wireless
network provided by an HCW unit.


The wireless network in the village wo
uld allow access to any vehicle equipped with an
OBWV unit, the purpose of the OBWV unit is to download the information stored on the
nurses computer via the HCW unit when it arrives in the village and when the vehicle
returns to the town of Sorata the OBW
V unit would upload the information to the health
offices database.
Thereby

acting as a courier of information between the town of Sorata
and the village of Quiabaya
[7]
.


By using this technique the system was able to provide the village with a limited data
connection where otherwise there was no connectivity at all.


6.2

Project Fontane


Experimental research has been conducted in the
field of remote community healthcare.
One of them being Project Fontane
[6]
.
Project Fontane is a collaborative research effort
by a consortium of numerous partners, among them include Deutsche Telekom, Hassno
-
Plattner
-
Institute, and some heart
specialists.
The project w
as launched to increase the
quality of medical care and to decrease the inconvenience of rural German citizens from
commuting vast distances for medical
check
-
ups
. Patients would be equipped

with monitoring smart boards (blood pressure meter, mobile electr
o
-
cardiographs etc.) and
the data would be sent to a tele
-
medical
centre

which would be monitored by medical
experts. A challenge which was experienced by the Fontane project was the high amount
of data being sent back to the tele
-
medical
centre
, since the

patients are being monitored
on a 24x7 basis, it also meant that there had to be medical experts available to diagnose
patients and store the data on the system. The Fontane project realized that due to the
highly variable amount of data that had to be tr
ansmitted and displayed to doctors. The
classification and prioritization of information could no longer be performed manually.
The process had to be automated to make the system scalable.

This was done by a combination of a expert systems as well as a mac
hine learning
algorithm, which were then able to discern which cases needed attention from medical
professionals. As the doctors were required to input data back into the system from their
responses to alerts, the systems were able to more accurately provi
de data and alerts to
medical professionals on standby.



7

Anticipated Outcomes:

This project has some architectural and interfacing challenges however, since the amount
of tools and community support for systems we plan to use is so large, we expect to
a
chieve a high percentage of our anticipate


7.1

System Outcomes



Accurate measurements of environmental parameters



User friendly interface for measuring
/viewing

of vitals parameters



Efficient use of energy for remote site, mobile devices and vehicle based
data courier unit



Data Courier unit and remote site should be able to operate for long periods without
maintenance.



Data is effectively distributed across systems to prevent bottlenecks and resources
hoarding.



Resources are effectively allocated to devices

with these allocations being dynamic based
on output received from grid members.



Able to correctly predict intervention based on sensor readings and historical data.



System is able to learn from feedback received from
usage

and able to adjust its analysis

of data.



Robust security implementation on WSBs and Android IOIOs



Simple web front end for medical professionals



7.2

Expected Impact

This project is mainly aimed at providing better healthcare for patients living in rural
areas. There will be a definite

impact on the quality of healthcare which these patients can
receive. The South African Medical authorities are making pushes towards becoming a
nation which stores patient history in a paperless way. Our system will show some key
insight into how the sto
rage of medical data could be conducted.


7.3

Security Outcomes

There is ofcourse a whole field of work dedicated to the systematic analysis and
evaluation of a secure communications network, for the purposes of not overscoping, We
will emply traditional
step by step method of evaluating a secure network.



8

Project Plan:

8.1

Risks

8.1.1 Hardware procurement delays

For the project certain devices are required, the devices include;



Environmental and biometric sensors



Alix smart boards



Android phone



Android

IOIO

Development of the system cannot begin until at least the Alix boards and Android IOIO
are acquired.

Severity:
High

Likelihood:
Medium

Mitigation strategy:
Ensure supervisor places the order for the devices as early as
possible. Source alternate devi
ces from personal or commercial sources.


8.1.2 Hardware

failure

Failure of any of the critical hardware components would cause a delay in the project
deliverables. Critical hardware components include Alix smart board and Android IOIO.

Severity:
High

Likelihood:
Low

Mitigation strategy:
During development care has to be taken while using the Android
IOIO as there is the risk of short
-
circuit due to its circuit board not being enclosed.


8.1.3 Failure to interface biometric sensors with mobile phone

The

interfacing between the various biometric sensors and the mobile phone requires the
Android IOIO. Time would have to be allocated to learn the functions of the Android
IOIO.

Severity:
High

Likelihood:
Low

Mitigation strategy:
Begin interfacing with the mo
st basic sensor type, the thermometer.
Doing this would allow time to learn the Android IOIO platform while completing one of
the sensing requirements.





8.1.4 Failure

to meet deliverable deadlines

During the project development there are sections of the

project that have to be handed in,
failure to meet deliverable deadlines may cause the following deliverables to be delayed
and result in project failure

Severity:
High

Likelihood:
Low

Mitigation strategy:
Having regular group meetings and with the projec
t supervisor to
ensure the project deadlines are adhered to. If the project is lagging behind schedule extra
work by all group members would be done to catch
-
up.







8.1.4 Failure to Establish Secure Communication Channels
Across

Platforms

There are
various communications links which have to communicate using encryption and
certificate signatures.

Severity:
High

Likelihood:
Low

Mitigation strategy:
Getting acquainted with the cryptographic libraries. Digital
Signatures can be implemented however crypt
ography should always be done using a
well
-
known library. Android has the javax.crypto library which a straightforward library
for encryption. All wireless communication will be using WPA2.


8.3

Resources

8.3.1

Security:



CryptoPP: Open Source Cryptography
Library for C++, Usable with any most Linux
distributions.



SSL: Internet communications cryptography standard.


8.3.2

Front End Application:



Django(Python Framework) or PHP



HTML/CSS/JavaScript


8.3.3

Sensors



Heart rate sensor



Thermometer



Gas sensor



Humidi
ty sensor



Technical documentation

8.3.4

Embedded computers



Android mobile phone or raspberry pi



Android IOIO



Alix smart board


Along with the actual hardware devices, the development environment software to write programs
would have to be acquired and
setup first. The Android SDK, Eclipse IDE and Java v7 SDK
comprise the development environment for the Android IOIO and Android mobile phone.


The Alix smart boards are capable of running a Linux distro. The distro that would be used for
this project is Op
enWrt. OpenWrt has all the necessary development software needed to write
programs for the Alix smart board.


8.4

Deliverables



Android application for measuring patient vitals



Distributed grid containing heterogeneous nodes



Data courier unit



e
-
Learning alg
orithm



Distributed database



Data warehouse



Web Front End



Robust secure

communications implementation



8.5

Milestones


Milestone

Duration

Due date

Project proposal presentations

3D

24th May

Revised project proposal finalized

20D

11th June

Project web
presence

5D

12th June

Feasibility demonstration

10D

13th June

Background/Theory chapter

7D

29th June

Design chapter begins

60D

29th August

First implementation

90D

19th September

Final prototype

99D

28th September

Outline of completed project

9D

10th

October

Draft of report

23D

24th October

Final report hand in

30D

31st October

Poster due

7D

3rd November

Web page due

3D

7th November

Project demonstrations

1D

8th November

Reflection paper

6D

11th November

Final project presentations

10D

17th
November




8.6

Work Allocation

8.6.1

Jarred Martin

Embedded system development



Android IOIO interfacing.



Android phone sensing application.



Communication between mobile device and remote site/data courier.

Remote site and vehicle based data courier



Alix

smart board environment sensing application.



Alix smart board traffic routing.



Wireless networking.


8.6.2

Taariq Mullins



Data warehouse



Data dissemination and aggregation



Data analysis



Grid monitoring



Temperature



Capacity



Throughput



Grid resource
management


8.6.3

Lutando Ngqakaza

Security and Data Privacy Responsibilities:



Network security for Delay Tolerant Networks.



Database security and account management.



Encryption and digital signatures for all network transactions and database queries.


Front End Dashboard Web Application:



Database access authentication.



Graphing.



Data display for medical professionals.



Environmental status indicator (uses environmental data).





9

Gantt chart:



























10

References


[1]

R. Buyya, C. S. Yeo, S. Venugopal, G. Computing, and S. Engineering, “Market
-
Oriented Cloud
Computing: Vision, Hype, and Reality for Delivering IT Services as Computing Utilities,” 1969.

[2]

A. Marinos and G
. Briscoe, “Community cloud computing Conference paper Community Cloud
Computing,” no. December, pp. 1
-
4, 2009.

[3]

T. Heinze, R. Wierschke, A. Schacht, and M. V. Löwis, “A Hybrid Artificial Intelligence System
for Assistance in Remote Monitoring of Heart
Patients,” pp. 413
-
420, 2011.

[4]

C. Neuhaus, R. Wierschke, M. V. L, and A. Polze, “Secure Cloud
-
based Medical Data Exchange.”

[5]

A. Schacht, R. Wierschke, M. Wolf, M. V. L, and A. Polze, “Live Streaming of Medical Data
-

The
Fontane Architecture for Remo
te Patient Monitoring and its Experimental Evaluation,” 2011.

[6]

A. Polze, P. Tr, U. Hentschel, and T. Heinze, “A scalable , self
-
adaptive architecture for remote
patient monitoring.”

[7]

M. J. Murillo and M. Aukin, “Application of Wireless Sensor Nodes t
o a Delay
-
Tolerant Health
and Environmental Data Communication System in Remote Communities,”
2011 IEEE Global
Humanitarian Technology Conference
, pp. 383
-
392, Oct. 2011.