Front-End Intelligence

FRONT END INTELLIGENCE FOR
LARGE-SCALE APPLICATION-
ORIENTED INTERNET OF THINGS
Presentedby:
Judy T Raj
CS7
ROLL NO:29

CONTENTS
• INTRODUCTION
• TRADITIONAL PARADIGM
• NEED FOR FRONT-END INTELLIGENCE
• BANDWIDTH DEMAND & DELAY INTOLERANCE
• DEVICE HETEROGENEITY

CONTENTS
• MOTIVATION FOR FRONT-END INTELLIGENCE
• QUALITATIVE RATIONALE
• ANALYTICAL RATIONALE
• COMPARATIVE SIMULATION ANALYSIS
• INTELLIGENT IOT DEVICES: CONNECTIVITY PERSPECTIVE
• CELLULAR NETWORK CONSIDERATIONS
• SPECTRUM MANAGEMENT
• D2D COMMUNICATIONS
• MULTIHOP NETWORKING
• INTELLIGENT IOT DEVICES: APPLICATION PERSPECTIVE
• INTELLIGENT IOT DEVICES: COLLABORATION PERSPECTIVE
• SOFTWARE DEVICE DRIVEN ARCHITECTURE
• CONCLUSION

INTRODUCTION
• What is IoT?
• Massive integration of electronic devices and other objects to collect &
exchange data.
• Enabling technology for multitude of applications in healthcare,
wearables, surveillance, home automation etc
• Aim of the paper
• Analyze the key performance matrices of back-end intelligence
• Advocate empowerment of front-end intelligence in IoT
• Discuss issues and challenges in empowering front-end IoT devices
• Analyze from three perspectives of IoT design : connectivity,
collaboration and application.

TRADITIONAL PARADIGM
• Centralized architecture
• Heavily reliant on the back-end core for all decision-making
processes
• All operational and management decisions related to front-end
resources are made at the back-end.
• Back-end or service platform in IoT is usually a cloud platform
• Inefficient in terms of latency, network traffic management,
computational capacity i.e. Instructions processed per unit time
and power consumption.

NEED FOR FRONT-END INTELLIGENCE
• This section outlines the design and operational
challenges brought forward by two trends associated
with the massive roll-out of IoT applications namely:
• Bandwidth Demand & Delay Intolerance
• Device Heterogeneity.

• BANDWIDTH DEMAND & DELAY
INTOLERANCE
• Growing trend of bandwidth hungry or
delay intolerant applications.
• Often observed in applications
demanding real-time transmission of
video streams.
• Examples include crowd management
applications, industrial processes, geo-
physical data acquisitions, cyber
physical systems etc.

• DEVICE HETEREOGENITY
• Refers to the wide spectrum of radio access technologies, network protocols,
capabilities, process power & communication technologies IoT devices support.
• No current standard for access protocols or messaging technologies for IoT.
• A Standardized homogenous solution is highly unlikely in future as well.
• Thus the need to cater to inherent technological diversity.
• Another disparity IoT gives rise to is multiple Administrative Domains (ADs).
• An AD refers to all IoT assets including front-end devices and back end platform
under the jurisdiction of a single entity.
• The coexistence of multiple ADs need to be accounted for.

MOTIVATION FOR FRONT-END INTELLIGENCE
• Interaction with other ADs cannot be performed if front-end
devices are not entitled to execute basic data gathering
operations about their neighbourhood of devices.
• The adverse implications of a highly-centralized paradigm can
be mainly assessed from two distinct perspectives:
networking efficiency and computational efficiency
• Advent of low-cost computationally efficient
processing units like microcontrollers provide
substantial motivation for front-end intelligent devices.

• Five key performance matrices can analyze the performance of IoT
devices.
• For back end intelligent systems, the corresponding values for each
term are:
• Energy: higher energy consumption
• Latency: longer trip times
• Throughput: lower throughput
• Scalability: Lesser scalability
• Reliability: Lower reliability

•
ANALYTICAL RATIONALE
• For large-scale multihop IoT setup where information must be exchanged for every
actuation.
• This cause lags and delays especially for FoT

• ANALYTICAL RATIONALE
• Back-end centric paradigm: the information collected from nearby
neighbours are forwarded to the back-end core to make the right
decision which is, in turn, sent to actuator to perform the required
action.
• Front-end empowered paradigm, the device is supported with
intelligence to take in-situ the decision based on the information
collected from nearby sensors that directly communicate with the
smart device.

• Paper investigated the impact of front end intelligence radio access procedure in LTE
networks and compared between three scenarios depending on the level of
collaboration.
• A Radio Access Technology or (RAT) is the underlying physical connection method for a
radio based communication network. As of 2013, many modern phones such as the
Nexus 4 or iPhone5 support several RATs in one device such as Bluetooth, Wi-Fi, and
3G, 4G or LTE.
• Long-Term Evolution (LTE) is a standard for high-speed wireless communication for
mobile phones and data terminals. It is based on the GSM/EDGE and UMTS/HSPA
network technologies, increasing the capacity and speed using a different radio
interface together with core network improvement

• Trad:all devices establish their connections with the node and complete their data
transfer independently.
• Collab: devices belonging to the same AD collaborate together to achieve data transfer.
• Social: devices socialize together and act as if they belong to a single virtual AD.

• DESIGN IMPERATIVE
• The paper aims to design a framework for the
design of IoT devices to empower front-end
intelligence.
• Can be achieved by viewing from three different
perspectives.
• Proposed framework supports diversity in
networking protocols.
• Comprehensive framework for design &
optimization of wide range of applications.
• Advocates socialization schemes enabling devices
of different ADs to share resources.

INTELLIGENT IOT DEVICES: CONNECTIVITY PERSPECTIVE
• Deals with main connectivity challenges i.e. Issues with cellular networks and spectrum management and
major promising directions that could cope with these issues.
CELLULAR NETWORK CONSIDERATIONS
• Based on simulation and analysis of actual traffic patterns of LTE Release 10 UL air interface in Saudi Arabia,
the paper concluded that a major capacity deficit occurs as applications become bandwidth hungry.
• As the number of connections/frame arrival increases, resource allocation becomes less efficient and physical random access
channel becomes congested in network entry procedure.
• Drop rate and channel access waiting time increase as connections increase.
SPECTRUM MANAGEMENT
• Effective power control strategies to maintain net interference levels within operationally acceptable ranges
need to be developed as the authors envision IoT communications outside licensed spectrum.

• PROMISING DIRECTIONS
• Further advancements in D2D communications and Multihop
networking to cope with the issues in connectivity.
• D2D communications: Is a technology component for LTE.
Enables direct communication between nearby mobile
equipments facilitate interoperability.
• Multihop Networks: use two or more wireless hops to convey
information from a source to a destination. Nodes in between
source and destination communicate using wireless channels.

D2D COMMUNICATION
• Can boost capacity in terms of
through put and number of
connections.
• Can be achieved by
ubiquitously available
technologies like WiFi with
effective scheduling to
manage human made and
M2M traffic.
• Access scheduling can cause
extensive interference in D2D
layer, battery depletion and
BW accounting and billing
challenges

MULTIHOP NETWORKING
• Another method to circumvent capacity deficit issues.
• Eg: smart grid, vehicle to vehicle communication
• Advantages: Standardized protocols such as THREAD are available
• Disadvantage: As the network builds in scale, protocol overhead occurs.
• Protocol overhead : metadata and network routing information sent by an application, which uses a
portion of the available bandwidth of a communications protocol.
• Solved to some extent by enabling IoT devices to make relaying decisions locally .
• IoT devices at a given hop extract position information encoded by the nodes of the previous
hop, and accordingly are able to locally make their routing decisions.

INTELLIGENT IOT DEVICES: APPLICATION ORIENTED PERSPECTIVE
• For optimal quality of service at end user level, few operational challenges exist:
• Battery management
• Device maintenance
• Software upgrades
• QoS requirements for each application might be different and affects the design.
• Paper recommends surveys to determine requirements before designing.
• Eg: Industrial locale has low throughput demand but is delay intolerant.
• Surveillance systems has high delay intolerance and bandwidth hunger.
• A descriptive framework to capture interactions between five KPMs for different
applications.

INTELLIGENT IOT DEVICES: COLLABORATIVE & SOCIAL
PERSPECTIVE
• Collaboration across ADs can meet the operational-application tradeoffs.
• Existing IoT applications using multiple ADs work in silos.
• Paper proposes a collaboration between different ADs to deliver best EUX within
constraints like QoS, privacy, security etc.
• Collaboration in IoT is mostly in three planes:
• Relaying & Routing information towards back-end server
• Intelligence and sensor measurements
• Computational resources
• Challenges: Different AD s have different QoS mandates
• Proposed solution: A Virtual marketplace for devices in different Ads to trade
resources

SOFTWARE DRIVEN DEVICE ARCHITECTURE
• Effectively integrates robust and communication platforms with provision
for different classes of traffic.
• Devising network parameters at device level rather than infrastructure level
helps cater to device heterogeneity.
• Objective is to develop software tools enabling front-end intelligence
combined with advantages of SDA like:
• Socialization
• Collaboration
• Dynamic return of operational parameters

IMPLEMENTATION
•End User Abstraction layer provides a GUI for users to provide inputs for customization of IoT device, translated to QoS.
•Parameterization Engine intelligently maps QoS requirements to the best operational profile i.e outputs a set of optimal values
for the device to observe.

CONCLUSION
The main contributions of the paper are summarized as follows:
• Demonstrates the advantage of front-end intelligence for large-scale application-oriented IoT
systems.
• Provides rather a novel model for optimizing the performance of application-oriented IoT in
light of connectivity constraints and collaboration potential.
• Identifies non-debatable shortcomings of some of the existing techniques and technologies in
the context making IoT devices intelligent.
• Proposes potential viable solutions while outlining open research directions.
In summary, this study defines a conceptual framework for future IoT devices enabling multiple
innovative features for the IoT platform administrator as well as the end user. These smart
IoT devices will have significant positive impacts on different domains allowing fast, reliable,
and intelligent management of diverse IoT-based applications.

Front-End Intelligence

Recommended

More Related Content

What's hot (20)

Viewers also liked (8)

Similar to Front-End Intelligence (20)

More from Judy T Raj (6)

Recently uploaded (19)

Front-End Intelligence

Editor's Notes