Professional Documents
Culture Documents
Research Directions
Davide Nardone
May, 2016
PARTHENOPE
UNIVERSITY
Outline
1. Introduction
2. IoT vision and contexts
3. The main IoT research directions:
A. Massive scaling
B. Architecture and Dependencies
C. Creating knowledge and Big Data
D. Robustness
E. Openness
F. Security
G. Privacy
H. Humans in the Loop
4. Conclusion
Machine
Learnig
Security Privacy
It’s clear the need for cooperation between the several communities for the
improvement and the resolution of various interesting research problem
regarding these environments.
Research Directions for IoT
IoT vision and contexts
• For many people, the concept of Smart City is no longer a vision so far
away.
• Many infrastructures are already equipped with no sensor and actuator
able to:
• Monitoring entire system and
• Interact with each other, in order to exchange information and so to
improve the quality of one or more services.
Examples
• Input: Biometric systems
• Output: Holographic display
? Challenges ✓ Benefits
• People / Society • Better and faster information
• Security / Privacy • Improving education
• Authority • Regulation / Legislation
• Standard / Policies • Life quality
• Resource Efficiency • Greater awareness
• Technology Architecture / Infrastructure • Increased production
• Costs • Wider experiences
• Pollution / Prevention disasters • Better decision making
• Innovation Management • Resource optimization
• Energy / Power Consumption • Human error removal
• Growth Data and analysis • More services
A. Massive scaling
B. Architecture and Dependencies
C. Creating knowledge and Big Data
D. Robustness
E. Openness
F. Security
G. Privacy
H. Humans in the Loop
MOBILE APPROACH
Advantages (+) Disadvantages (-)
• Unlimited application development. • Interference problems due the sharing of
sensors and systems implementation
among multiple applications.
• Automatic controls for the execution of an • Dependence of IT methodologies based
application on a specific platform. on the assumption about the environment,
the hardware platform, naming, control
and various semantic devices.
Let’s suppose the integration of different systems responsible for the electricity
management (control thermostats, windows, doors, etc.) and the health care
(ECG measurement, temperature, sleep, activities, etc.).
Advantages
• Information sharing: this would allow the electricity management systems to
regulate the temperature of the rooms according to the physiological status of
the residents.
• Integration between systems: this would avoid unpleasant situations; for
instance the system will deactivate medical devices to save power while they
are used as suggested by the health care system.
• Sharing sensors and actuators: this would reduce the costs of
implementing/distribution, improving the aesthetics of the rooms and reducing
the flow of content.
Example of interference
• The health care system could detect depression and decide to turn on all the
lights. On the other hand, the electricity management system may decide to
turn off the lights due to the absence of motion.
Possible solutions
• Associate a confidence level (in the form of probability) to the
information derived from the date.
• Using Fuzzy Logic [4].
• Minimizing the number of false negatives and false positives.
For example, how do you deal with the situation where the fridge and coffee
maker are equipped with hidden Wi-Fi access and spammers?
According to the experts, the problem of data security has two major
aspects:
1. Confidentiality
2. Protection
Example
One of the most important transportation companies in the world, UPS, uses
sensors in its vehicle to improve delivery performance and reduce coast.
• These data help UPS out in reducing fuel consumption, harmful emissions
and waste time.
47%
53%
87%
Challenges
The deterioration of these conditions cause the break up of the clocks that:
• Causes the nodes to have different time shift, that leads to failures
within the system.
Solutions
Synchronization techniques are used in this context [7].
The higher the entropy, the greater the disorder and less energy is available in the
system to do useful things.
Problems
• IoT products are often sold with old
embedded operating systems and software
products.
• Many of these systems/software (designed
to be safe) are still vulnerable.
Given the particular nature of these devices, IoT has the following
challenges to face:
1. Critical functions
2. Replication
3. Difficulty in repairing
4. Security assumptions
5. Long life cycle
6. etc.
Solutions used for standard computer do not solve the IoT security
issues of embedded devices, due to the their heterogeneity.
Secure code updates Safe methods for updating the code, bugging fixes, security
patches, etc. (use the signed code).
Data security Prevent unauthorized access to the device, encrypting the data
storage and / or communication.
Secure communication Communication to and from the devices must be secured using
encrypted protocols (SSH, SSL, etc.).
Protection against cyber attacks Integrated firewall to restrict communications to only trusted host,
blocking hackers before they can launch attacks.
Intrusion detection and security Detection and communication of disabled login attempts and other
monitoring potential malicious activity.
Tamper detection on device New processors / cards that have capacity of device tampering
detection.
Self Healing Methods for "healing" from cyber attacks (e.g., code updates).
Solutions
1. Specific privacy policies for each domain / system and
2. Privacy reinforcement in the IoT infrastructure of the IoT
applications.
IoT paradigm
1. Expression of users' requests for accessing to the data and the
policies they have adopted.
2. Assessment of the requests to ensure whether or not their
permission.
Example:
• Optimization of human behavior based on reports of accidents and incidents
that occur during operation of an electric power system:
• Components Models of Emotion (CME) for observing, recording, and
analyzing the emotional component of the operator's behavior.
• Simulation of the dynamic behavior of an operator in performing
operations in a context that leads to an error.
Expectations
It is hoped that there is more cooperation between the research
communities in order to:
• Solve the myriad of problems as soon as possible;
• Avoid reinventing the wheel when a particular community solves a problem.