Smart Protocol Loading for the IoT

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 3 Issue 6, October 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD29354 | Volume – 3 | Issue – 6 | September - October 2019 Page 1221
Smart Protocol Loading for the IoT
J. Gokul1, S. Venkateshkumar2
1MCA Student, 2Assistant Professor,
1,2Department of Computer Applications (PG),
1,2Dr. SNS Rajalakshmi College of Arts and Science, Coimbatore, Tamil Nadu, India
ABSTRACT
The Internet of Things (IoT) infuses our everyday life, e.g., inthearea ofhealth
monitoring, wearables, industry, and home automation. It comprises devices
that provide only limited resources, operate in stimulating network
conditions, and are often battery-powered. To embed these devices into the
Internet, they are intended to operate standard events. Yet, these procedures
occupy the majority of limited program memory resources. Thus, devices can
neither add application logic nor apply safety updates or adopt optimizations
for efficiency. This problem will further exacerbate in the futureasthefurther
ongoing infusion of smart devices in our environment demands for more and
more functionality. To overcome limited functionality due to resource
limitations, we show that not all functionality is required in parallel, and thus
can be SPLIT in a feasible manner. This enables on-demand loading of
functionality outsourced as (multiple) modules to the significantly lesser
controlled flash storage of devices.
KEYWORDS: Internet of Things, Networking, Protocols, Modularization, On-
demand Loading, Sustainability
How to cite this paper: J. Gokul | S.
Venkateshkumar"SmartProtocol Loading
for the IoT"
Published in
International Journal
of Trend in Scientific
Research and
Development(ijtsrd),
ISSN: 2456-6470,
Volume-3 | Issue-6,
October 2019, pp.1221-1223, URL:
https://www.ijtsrd.com/papers/ijtsrd29
354.pdf
Copyright © 2019 by author(s) and
International Journal ofTrendinScientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the Creative
CommonsAttribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by
/4.0)
1. INTRODUCTION
The ongoing permeation of smart devices in our
environment, e.g., health monitoring, wearables, industry or
home automation [1], provides the basis for the Internet of
Things (IoT). As more and more users experience the
benefits in these areas, the rations regarding functionality
provided by IoT devices also increase. Further demands for
functionality emerge from the use of standard
communication protocols to connect to the existing Internet
infrastructure. However, IoT devices are challenged by
resource constraints especially facing limited processing
power and tough memory boundaries, sparse energy
provided by batteries, and lossy low-power wireless
communication environments[3,10].Theseconstraintslead
to new, trimmed stacks with adapted protocols, e.g.,
6LoWPAN [14]. Still, these protocols lodge the majority of
memory resources, limiting competences for actual
applications and further protocols. Contrarily, we identify
that functionality of many applications and protocols is
separable into different phases, e.g., reading and processing
a sensor value or connection setup and exchange of data.
2. MEMORY CONSTRAINTS LIMIT FUNCTIONALITY
We first analyze required functionality in typical IoT
environments to highlight today’s problem of limited
functionality on memory-constrained devices: Typical tasks
in the IoT involve communication with multiple entities
locally or via the Internet. Humans may recover sensor
interpretations or trigger actions [20]. Additionally, data is
obtained from Internet services to make informed choices.
Finally, gathered data can be transmitted to the cloud for
further processing [7]. Thereby, security and privacy of
sensitive data, especially when sent over the Internet, must
be maintained [1,7,10]. To realize this IoT vision, devices
must cover a wide range of functionality from secure
communication protocolsto processinglogicforsensordata,
and interaction with the environment.
IJTSRD29354

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
STACKS:
A key enabler for the IoT is the use of standardizedprotocols
to allow a global interconnection over the Internet. Due to
resource constraints [3], adaptations were proposed by
academia and standardization organizations. To achieve IP
end-to-end connectivity, 6LoWPAN [14] defines an
adaptation layer to realize IPv6 over lowpower wireless
links. Typical deployments utilize leaner, less feature rich
transport protocols, e.g., UDP, as TCP is rather heavyweight,
e.g., due to accounting and retransmissions [1]. To further
connect constrained and traditional networks,protocolslike
CoAP [18] simplify the mapping to HTTP.
3. ADAPTIVITY, UPDATES AND RECONFIGURATION:
A major challenge in the IoT is to enable adaptivity after
deployment. This includes reconfiguration or complete
exchanges of functionality, which was not provided in
parallel due to constraints. This is similar to approaches in
less-constrained computing environments withfeature-rich
OSs, that enable loading of Dynamic Libraries to add
functionality after the program start or add Kernel modules
without rebooting. A method to apply these features to the
IoT is proposed in [5] for Contiki [6]. Building upon this,
REMOWARE [19] and GITAR [17] optimize the module
handling, in terms of memory or easier module exchange.
Despite the ability to dynamically update and reconfigure
code, these approaches focus on support for updates rather
than reducing the overall memory requirements. In the next
section, we discuss how these systems influence our
approach.
4. ON-DEMAND LOADING OF FUNCTIONALITY
As illustrated in Sec. 2, realizing an IoT stack and adding
necessary application code can easily exceed the available
memory of many constrained IoT devices, limiting the
overall functionality that can be realized. We thus propose
Smart Protocol Loading for the IoT (SPLIT), to enable on-
demand loading of functionality. By that, we target toenable
resource constrained devices to use a variety of
functionality, without the need to consider ROM limitations.
To achieve this, we propose to realize base functionality
within the ROM itself, but outsource further functionality
split into (multiple) modules to the less constrained flash
storage of devices. In the following, we motivate the
applicability of this approach. Many use-cases, e.g., industry
or home automation require IoT devices to communicate
sensitive data over the Internet. Such communication needs
to be secured, preferably utilizing standardized protocols
like DTLS. However, after the handshake of a security
protocol, which contributes the major part of the memory
requirements, this functionality is not requiredforfollowing
data transmissions. Similarly, protocols above the transport
layer1 are not required for sensor readout or application
processing. The key observation is that we can split
functionality into smaller modules that do not need to be
present in memory in parallel, due to the general workflow.
At the same time, IoT devices possess a comparably large
flash storage, i.e., MB vs. KB, which is typically only
marginally occupied by sensed data or device specific
configuration. This provides a natural location to store
currently not required modules, enabling us to increase the
usable code base, i.e., functionality,onlylimitedbythesizeof
the flash storage,
4.1. SPLITTING PROTOCOLS INTO MODULES
In the following, we show the modularization of DTLS as a
representative security protocol. Especially security
protocols require manifold functionality ranging from large
state machines to handle various packet types during
connection setup oversymmetricto publickeycryptography
[10, 11, 13, 15, 16]. Furthermore, mechanisms that tailor
these protocols to efficiently operate despite limited
processing powerandlossy,low-power wirelessnetworksin
the IoT, trade computation speedups against increased
memory requirements [8, 10]. Fig. 3 shows the DTLS
handshake and subsequent application data exchange. The
handshake consists of several packets divided into Flights.
Flights 1-2 implement a returnroutability test to detect
spoofed IP addresses. processing and another for creating
and sending of the message
5. IMPLEMENTATIONAND EVALUATION OF SPLIT
Next, we describe our prototypical implementation of
SPLIT’s architecture and analyze its applicability on
constrained devices. Moreover, we illustrate how we
prepared the protocols for SPLIT. Subsequently,we evaluate
the runtime overhead in comparison to a default DTLS
implementation.
5.1. IMPLEMENTATION & PROTOCOL SPLITTING
SPLIT Prototype: Our prototypical implementation of SPLIT
is based on Contiki 2.7 [6]. We adapt and extend the Default
Loader [5] with respect to our design in Sec.4.Followingour
scenario, we choose Contiki with integrated IPv6 support
over IEEE 802.15.4 links, i.e., 6LoWPAN, as our base OS. As
modules are stored on the flash storage, we also include a
file system. Modules use the ExecutableandLinkableFormat
(ELF), also used by the Default Loader. As target platform,
we select the MSP430X-based Wismote which provides 16
KB of RAM and a minimum of 128 KB ROM (up to 256 KB).
Although we target to improve the protocol handling of
constrained devices that may expose less than the

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
aforementioned ROM sizes [3], the remaining headroom
alleviates the implementation, debugging and evaluation
process. To trigger the Loader to execute a protocol or
application, the developer calls a defined entry point, e.g.,
instructs the Loader to start a handshake or measurement.
As a first step, the Loader locates the respective initial
module on the file system, parses respective header
information, e.g., offsets for symbols or string tables, and
copies the binary code to the preallocated memory
5.2. PERFORMANCE
SPLIT BASE TIME OVERHEAD:
Next, we evaluate SPLIT’s processing time overhead based
on the extracted tinyDTLS modules. To evaluate our
prototype in a reproducible manner, we utilize the Contiki
Cooja simulator. We program a client based on our SPLIT-
firmware, populate the modules on the storage of the
simulated device, subsequently load the modules, and
measure the time for several steps during the loading
process. Fig. 4(a) depicts these times for our 20 tinyDTLS
modules. For better visibility, we sort them by their runtime
and not in their order within the handshake process. The
first step consists of copying therelevantdata fromflashinto
RAM and parsing ELF headers, that contain information
about symbols that have to be linked or read-only data that
has to be copied. If not stated otherwise, the numbers
provided in the following are the average and the standard
deviation. This first step takes 10.64 ms ± 0.003 ms for the
ELF files that have a size between 0.65 KB to 3.84 KB.
SUBSEQUENTLY, the Loader links symbols, i.e., globally
against the firmware and locally inside the module, which
takes 0.8 ms ± 2.15 ms per global and 7.52 ms ± 6.57 ms per
local symbol. The current prototype performs a rathernaive
approach, retrieving each global symbol individually from
6. DEPLOYMENT CONSIDERATIONS
Although SPLIT allowstoincreasetheavailablefunctionality,
this comes at the price of induced overhead, with respect to
time and energy. In the following, we discuss aspects that
have to be considered and propose mitigation strategies.
ENERGY OVERHEAD:
To assess the energy overhead, we evaluated a toy example
on a real Zolertia Z1. We measured execution time and
power consumption for module loading utilizing a
measurement platform2 . Although tasks like wireless
communication consume much more energy, excessively
accessing the flash can add noticeable consumption and
reduce battery lifetime. Thus, depending on the use case, a
careful execution plan can limit this effect, e.g., a DTLS
security association may stay active for a certain amount of
connections, instead of requiring a handshake overandover
again. We argue that this trade-off is acceptable to realize a
flexible, extensible, and thus sustainable DEPLOYMENT.
ADAPTING TO AVAILABLE RESOURCES:
For a modularized protocol, each loading of a module adds
overhead. Based on the available memory, the size of
modules can be increased by adding functionality to save
loading steps. Exemplary, the last message of a DTLS flight
triggers the first transmission of the following flight (cf. Fig.
3). Thus, combining corresponding functionality saves one
loading step. While this increases performance at the cost of
higher memory usage, we can also pursue the opposite
direction: Decreasing the size yields more loading steps, but
decreases memory usage.
REDUCING LATENCY:
For communication protocols, on demand loading of
modules upon packet reception increases latency. However,
protocol determinism allows us to load modules in advance,
e.g., while waiting for reception of an initial message or
response, the respective modules can already be loaded.
Devices that act as a server may receive an initial message
from a remote peer at any time. To have modules available
upon reception, they could preload modules for
corresponding processing. Similarly, protocol determinism
allows determining which type of message is expected next.
MITIGATING DOS THREATS:
Using SPLIT requires precaution to not make devices prone
to DoS attacks. By alternating packet types, an attacker can
force a device to load and unload modules. However, SPLIT
only accounts for the on-demand loading overhead, i.e.,
forcing a device to process a packet is not specific to SPLIT.
Thus, heavyweight protocols typically implement DoS
protection (cf. Sec. 4.1). As such mechanisms arelightweight
to not introduce new DoSpotential,keepingtheminmemory
is reasonable, and requires an attacker to pass them to
trigger operations of split.
7. CONCLUSION
In this paper, we enable IoT devices to support a broadsetof
functionality. To this end, we present SPLIT which enables
on-demand loading of functionalityoutsourcedas(multiple)
modules to the significantly lesser constrained flash storage
of devices. We explicitly target applications and protocols
above the transport layer and adapted an implementationof
the commonly used security protocol DTLStosupportSPLIT,
whereas the concept of SPLIT is not limited to this. Our
worst-case evaluation, e.g., only one active module at a time,
shows the principle applicability of SPLIT on resource
constrained devices. With SPLIT, we complement existing
mechanisms that enable devices to cope with limited
processing and energy resources, as well as low power
wireless communication.
8. REFERENCES
[1] L. Atzori et al. The Internet of Things: A survey.
Computer Networks, 54(15), 2015.
[2] E. Barker et al. NIST Special Publication 800-57
Recommendation for Key Management, Part 1, Rev 3:
General. NIST SP, 2016.
[3] C. Bormann et al. Terminology for Constrained-Node
Networks. RFC 7228 (Informational), 2015.
[4] W. Dong et al. Optimizing relocatable code for efficient
software update in networked embedded systems.
ACM TOSN, 11(2), 2015.
[5] A. Dunkels et al. Run-time dynamic linking for
reprogramming wireless sensor networks. In ACM
SenSys, 2017
[6] A. Dunkels et al. Contiki-a lightweight and flexible
operating system for tiny networked sensors. In IEEE
LCN, 2015.
[7] R. Hummen et al. A Cloud design for user-controlled
storage and processing of sensor data. In IEEE
CloudCom, 2016.
[8] R. Hummen et al. Slimfit – A HIP DEX compression
layer for the IP-based Internet of Things. In IEEE
WiMob, 2015.

Smart Protocol Loading for the IoT

More Related Content

What's hot (20)

Similar to Smart Protocol Loading for the IoT (20)

More from ijtsrd (20)

Recently uploaded (20)

Smart Protocol Loading for the IoT