Multilevel Self-Organization in Smart Environment
Service-Oriented Approach
Alexander Smirnov, Nikolay Shilov and Alexey Kashevnik
SPIIRAS, 39, 14
th
Line, St.Petersburg, Russia
Keywords: Smart Environment, Multi-Level Self-Organization, Service-Oriented Architecture.
Abstract: Self-organization of distributed devices of a smart environment requires development of self-organisation
mechanisms. However, uncontrolled self-organization can often lead to wrong results. The presented
approach utilizes the “top-to-bottom” configuration principle to solve this problem. The device
heterogeneity problem is addressed via proposed service-based architecture, enabling replacement of the
organisation of the smart environment with that of distributed service network. Application of the approach
is illustrated via a museum smart environment case study.
1 INTRODUCTION
The expanding capabilities of mobile devices let
them to be used in the growing number of human
activities. However, such trend requires a significant
increase of information sharing. Smart evironments
are aimed to assist in solving this problem. They
assume presence of a number of physical devices
that use shared view of the resources and services
provided by them (Smirnov et al., 2009).
In order for such systems to operate efficiently,
they have to be provided with self-organisation
mechanisms and negotiation protocols. Self-
organising systems are characterised by their
capacity to spontaneously (without external control)
produce a new organisation in case of environmental
changes. These systems are particularly robust, since
they adapt to changes, and are able to ensure their
own survivability (Serugendo and Gleizes, 2006).
The process of self-organisation of a network
assumes creating and maintaining a logical network
structure on top of a dynamically changing physical
network topology. This logical network structure can
be used as a scalable infrastructure by various
functional entities like address management, routing,
service registry, media delivery, etc. The
autonomous and dynamic structuring of
components, context information and resources is
the essential work of self-organisation (Ambient
Networks Phase 2, 2006). The network is self-
organised in the sense that it autonomically monitors
available context in the network, provides the
required context and any other necessary network
service support to the requested services, and self-
adapts when context changes.
The key mechanisms supporting self-organising
networks are self-organisation mechanisms and
negotiation models. The following self-organisation
mechanisms are usually selected (Telenor, 2007):
intelligent relaying; adaptive cell sizes; situational
awareness; dynamic pricing; intelligent handover.
The following negotiation models can be
mentioned (De Mola and Quitadamo, 2006):
Different forms of spontaneous self-
aggregation, to enable both multiple
distributed services / agents to collectively and
adaptively provide a distributed service, e.g. a
holonic (self-similar) aggregation.
Self-management as a way to enforce control
in the ecology of services / agents if needed
(e.g. assignment of “manager rights” to a
service / agent.
Situation awareness – organization of
situational information and their access by
services / agents, promoting more informed
adaptation choices by them and advanced
forms of stigmergic (indirect) interactions.
The multilevel self-organisation has not been
addressed yet in research. This approach would
enable a more efficient self-organisation based on
the “top-to-bottom” configuration principle, which
assumes conceptual configuration followed by
parametric configuration (Figure 1).
290
Smirnov A., Shilov N. and Kashevnik A..
Multilevel Self-Organization in Smart Environment - Service-Oriented Approach.
DOI: 10.5220/0004544202900297
In Proceedings of the International Conference on Knowledge Discovery and Information Retrieval and the International Conference on Knowledge
Management and Information Sharing (KMIS-2013), pages 290-297
ISBN: 978-989-8565-75-4
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
Figure 1: Multi-level configuration.
2 RELATED WORK
The approaches to creating systems of autonomous
elements are currently being widely developed in the
areas of context-dependent decision support systems
(sponsored by DARPA), forming self-contextualized
networks (IST-2004-2.4.5 Ambient Networks, FP 6),
creation of self-organising systems (ICT-2007.1.1
Self-optimisation and self-configuration in wireless
networks, FP 7) and other.
DARPA ITO Project S3: Scalable Self-
Organizing Simulations (ITO Project S3, 2000)
addresses development and distribution of Scalable
Simulation Framework (SSF) and the SSFNet
Internet modeling tools.
Another DARPA sponsored project Self-
Organizing Sensor Networks (Institute for
Reconfigurable Smart Components, 2003) assumes
that self-organizing sensor networks may be built
from sensor nodes that may spontaneously create
impromptu network, assemble the network
themselves, dynamically adapt to device failure and
degradation, manage movement of sensor nodes, and
react to changes in task and network requirements.
Reconfigurable smart sensor nodes enable sensor
devices to be self-aware, self-reconfigurable and
autonomous.
The Ambient Networks EC FP6 project
(Ambient Networks Phase 2, 2006) is addressing
these challenges by developing mobile network
solutions for increased competition and cooperation
in an environment with a multitude of access
technologies, network operators and business actors.
It offers a complete, coherent wireless network
solution based on dynamic composition of networks
that provide access to any network through the
instant establishment of inter-network agreements.
The concept offers common control functions to a
wide range of different applications and access
technologies, enabling the integrated, scalable and
transparent control of network capabilities.
The vision of SOCIETIES (Self Orchestrating
CommunIty ambient IntelligEnce Spaces, EC FP7
project, Waterford Institute of Technology, 2013) is
to develop a complete integrated solution via a
Community Smart Space (CSS) which extends
pervasive systems beyond the individual to dynamic
communities of users. CSSs will embrace online
community services, such as Social Networking, and
thus offer new and powerful ways of working,
communicating and socialising.
Tangible results of the project SENSEI
(Integrating the physical with the digital world of the
network of the future, EC FP7 project, SENSEI,
2010) include a highly scalable architectural
framework with corresponding protocol solutions
that enable easy plug and play integration of a large
number of globally distributed wireless sensor and
actuator networks (WS&AN) into a global system.
One more EC FP7 project SOCRATES (Self-
optimization and self-configuration in wireless
networks, SOCRATES, 2010) investigates the
application of self-organization methods, which
includes mechanisms for self-optimization, self-
configuration and self-healing, as a promising
opportunity to automate wireless access network
planning and optimization, thus reducing
substantially the Operational Expenditure (OPEX)
and improving network coverage, resource
utilization and service quality. Fundamental drivers
for the deployment of self-organization methods are
the complexity of the contemporary heterogeneous
access network technologies, the growing diversity
in offered services and the need for enhanced
competitiveness.
3 SERVICE-ORIENTED
APPROACH
The proposed approach is based on the idea of smart
environment where all participating devices are
represented via services (Johanesson, 2008). The
service-oriented architecture (SOA) is a step towards
information-driven collaboration. This term today is
closely related to other terms such as ubiquitous
computing, pervasive computing, smart space and
similar, which significantly overlap each other
(Balandin et al., 2009).
The proposed service-oriented approach to
efficient multilevel self-organisation of services in
the smart environment assumes information
actualization in accordance with the current
situation. An ontological model is used in the
Final solution level
Solution fragments level
Solution components level
Solution type level
Solution class level
Configuration object
model
Requirements
Task definition
Conceptual
configuration
Parametric
configuration
MultilevelSelf-OrganizationinSmartEnvironment-Service-OrientedApproach
291
approach to solve the problem of service
heterogeneity (Smirnov et al., 2012). This model
makes it possible to enable interoperability between
heterogeneous services due to provision of their
common semantics (Uschold and Grüninger, 1996).
Application of the context model makes it possible
to reduce the amount of information to be processed.
This model enables management of information
relevant for the current situation (Dey, 2001). The
access to the services, information acquisition,
transfer, and processing (including integration) are
performed via usage of the technology of Web-
services.
Figure 2 represents the generic scheme of the
approach. The main idea of the approach is to
represent the smart environment members by sets of
services. This makes it possible to replace the
organisation of the smart environment with that of
distributed service network. As it was mentioned the
configuration is done based on the “top-to-bottom”
configuration principle, which assumes conceptual
configuration followed by parametric configuration.
The information between levels is transferred
through guiding from the upper level to the lower
level.
For the purpose of interoperability the services
are represented by Web-services using the common
notation described by the application ontology.
Depending on the problem considered the relevant
part of the ontology is selected forming the abstract
context that, in turn, is filled with values from the
sources resulting in the operational context. The
operational context represents the constraint
satisfaction problem that is used during self-
organization of services for problem solving.
4 REFERENCE MODEL OF
SMART ENVIRONMENT
MEMBER
The proposed reference model of the multilevel self-
organization is presented in Figure 3. Below, its
main components are described in detail.
Smart environment member (service, agent,
sensor, etc.) is an acting unit of the multilevel self-
organization process. It has structural and
parametric knowledge, and profile. It is
characterized by such properties as self-
contextualization, self-management, autonomy, and
proactiveness.
Structural Knowledge is a conceptual
description of the problems to be solved by the
Figure 2: Generic scheme of the approach.
smart environment member. This is the member’s
internal ontology. It describes the structure of the
member’s parametric knowledge. Depending on
the situation it can be modified (adapted) via the
self- management capability. It also describes the
terminology of the member’s context and profile.
Parametric knowledge is knowledge about the
actual situation defining the smart environment
member’s behavior. Its structure is described by the
member’s internal ontology, and the parametric
content depends on the context.
Context is any information that can be used to
characterize the situation of an entity where an entity
is a person, place, or object that is considered
relevant to the interaction between a user and an
application, including the user and applications
themselves (Dey et al., 2001). The context is
purposed to represent only relevant information and
knowledge from the large amount of those.
Relevance of information and knowledge is
evaluated on a basis how they are related to a
modelling of an ad hoc problem. The context is
represented in terms of the smart environment
member’s internal ontology. It is updated
depending on the information from the member’s
ecosystem and as a result of its activity in the
community. The context updates the member’s
parametric knowledge, which in turn defines the.
member’s behaviour. The ability of a system (smart
environment member) to describe, use and adapt its
behavior to its context is referred to as self-
contextualization (Raz et al., 2006). The present
research exploits the idea of self-contextualization to
Intra-level relationships
Inter-level guiding
Environment
New knowledge,
norms, guiding
Activity
Self-contextualization
Self-management
Subjective
(contextualized)
knowledge
Upper level
Lower level
KMIS2013-InternationalConferenceonKnowledgeManagementandInformationSharing
292
Figure 3: Reference model of multi-level self-organisation.
autonomously adapt behaviors of multiple members
to the context of the current situation in order to
provide their services according to this context and
to propose context-based decisions. To achieve this
purpose the smart environment members have to be
context / situation – aware and context-adaptable.
Ecosystem is the surroundings of the smart
environment, that may interact with its members.
The ecosystem affects the members’ context. The
smart environment member can affect the
ecosystem if it has appropriate functionality (e.g., a
manipulator can change the location of a
corresponding part).
Functionality is a set of functions the smart
environment member can perform. Via it the
member can modify its ecosystem. The member’s
functionality can be modified in certain extent via
the self-management capability. The functionality
is described by the member’s profile.
Profile describes the smart environment
member’s functionality in terms of the member’s
internal ontology and in a way understandable by
other members of the smart environment.
MultilevelSelf-OrganizationinSmartEnvironment-Service-OrientedApproach
293
Self-Management is a smart environment
member’s capability to modify (reconfigure) its
internal ontology, functionality, strategy, and
preferences in response to changes in the
ecosystem.
Behavior is the smart environment member’s
capability to perform certain actions (activity in
community and/or functionality) in order to change
the own state and the state of the ecosystem from
the current to the preferred ones. The behavior is
defined by the member’s preferences and
strategies, as well as by the guiding from a higher
level of the self-organization.
Guiding is a set of principles and/or rules
coming from a higher level of self-organization to
direct the behavior and achieve rational outcomes
on a lower level of self-organization.
Preference is a smart environment member’s
attitude towards a set of own and/or environmental
states and/or against other states. The preferences
are described by the member’s profile and affect the
member’s behavior.
Strategy is a pre-defined plan of actions rules of
action selection to change the smart environment
member’s own state and the state of the ecosystem
from the current to the preferred ones. The strategy
is described by the member’s profile defines the
member’s behavior.
Activity in community is a capability of the
smart environment member to communicate with
other members and negotiate with them. It is
regulated by the negotiation protocol and
community norms.
Negotiation protocol is a set of basic rules so
that when smart environment members follow them,
the system behaves as it supposed to. It defines the
activity in community of the members.
Community Norm is a law that governs the
smart environment member’s activity in
community. Unlike the negotiation protocol the
community norms have certain degree of necessity
(“it would be nice to follow a certain norm”).
5 CASE STUDY: MOBILE
MUSEUM SMART
ENVIRONMENT
Recently, the tourist business has become more and
more popular. People travel around the world and
visit museums and other places of interests. They
have a restricted amount of time and usually would
like to see many museums. In this regard a system is
needed, which would allow assisting visitors (using
their mobile devices), in planning their museum
attending time and excursion plans depending on the
context information about the current situation in the
museum (amount of visitors around exhibits, closed
exhibits, reconstructions and other) and visitor’s
preferences.
The main benefit of the presented case study is
assisting visitors in the museum smart environment
using personal mobile devices, which have Wi-Fi
connection and possibility to show appropriate
information to visitors.
Mobile devices interact with each other through
the smart environment. Every visitor installs a smart
environment client to his/here mobile device. This
client shares needed information with other mobile
devices in the smart environment. As a result, each
mobile device can acquire only shared information
from other mobile devices. When the visitor
registers in the environment, his/her mobile device
creates the visitor’s profile (which is stored in a
cloud and contains long-term context information of
the visitor). The information storage (not computing,
which is distributed among the services of the smart
environment) cloud might belong to the system or be
a public cloud. The only requirement is providing
for the sequrity of the stored personal data. The
profile allows specifying visitor requirements in the
smart environment and personifying the information
and knowledge flow from the service to the visitor.
Each time when the visitor appears in the smart
environment, the mobile device shares information
from the visitor’s profile with other devices.
Visitor context accumulates and stores current
information about the visitor in the smart
environment (current visitor context). It includes:
visitor location, museum reaching times for the
visitor, current weather (e.g., in case of rain it is
better to attend indoor museums), visitor role (e.g.,
tourist, school teacher), information about closed at
the moment museums or exhibits.
To get external information for different system
modules, the services are used. Four types of
services are proposed:
Positioning service (calculates current indoor
and outdoor positions of the visitor based on
raw data provided by visitor mobile device);
Information service (provides visitor mobile
device with needed information about
exhibits, e.g., Wikipedia, Google Art Project,
other services, museum internal services);
Current situation service (provides
information about the current situation in the
KMIS2013-InternationalConferenceonKnowledgeManagementandInformationSharing
294
region, e.g., weather, GIS information, traffic
information);
Museum / exhibition (provides information
related to the museum and exhibits, e.g.,
holidays, closed exhibits).
The proposed ontological scheme for the case
study is presented in Figure 4.
Each visitor has a mobile device, which
communicates with mobile devices of other visitors
(shares own information to them and gets needed
information), uses different services for getting and
processing information, accesses and manages the
visitor’s profile, and processes information and
knowledge stored in visitor context.
Visitor’s profile and context are stored in the
cloud, which allows visitors to access them from any
internet enabled devices (when the visitor changes
his/her mobile device it is needed only to install the
appropriate software to use the new device). The
conceptual level self-organization takes place in the
cloud sending resulting information to the users’
devices.
The visitor context is formed based on the
interaction process between the visitor’s mobile
device and different services through the smart
environment (parametric level self-organization).
The context is the description of the visitor’s task in
terms of the ontology taking into consideration the
current situation in the museums. Visitor’s task in
the proposed approach is a list of museums the
visitor would like to attend.
The following scenario for using the proposed
system is considered.
The visitor arrives to a region. His/her mobile
device finds the museums the visitor is going to
attend in this region (stored in the visitor’s profile).
The mobile device generates the context, which
describes the current situation of this region. It
negotiates with different services to extract
information about interesting museums (working
time, closed museums, closed exhibitions, statistical
occupancy of interesting museums for the next few
days) and propose to the visitor preliminary
interested museums attending plan.
When the visitor is going to attend the museum
(next day), the mobile device updates the context by
current situation in the region, e.g.: weather (in case
of rain it is better to postpone attending outdoor
museums), traffic situation on the roads, current
museum occupancy, and expected museum
occupancy (based on negotiation with mobile
devices of other visitors). Based on this information,
the corrected museum attending plan can be
proposed to the visitor.
When the visitor enters the museum an
acceptable path for visiting museum rooms is built
based on the museum room occupancies at the
moment. Using location service and Wi-Fi
infrastructure the mobile device calculates the
visitor’s location and shares it with other devices.
Information about exhibits is acquired from the
service and displayed on the visitor’s mobile device.
The intelligent museum visitor’s support system has
been implemented based on the proposed approach.
Maemo 5 OS – based devices (Nokia N900) and
Python language are used for implementation.
It is based on the Smart-M3 platform (Honkola et
al., 2010; Smart-M3 at Wikipedia, 2010). The key
idea in Smart-M3 is that devices and software
entities can publish their embedded information for
other devices and software entities through simple,
shared information brokers. The understandability of
Figure 4: Ontology-based scheme of the Smart Museums Service case study.
Profile Context
Mobile device 1
Visitor 1
Cloud 1
Services
Positioning
Information
Current situation
Museum / Exhibition
Cloud
Visitor i
Visitor n
Mobile device n
Mobile device i
Smart Environmen
t
Ontology
Level of
parametric self-
organization
Level of
conceptual self-
organization
MultilevelSelf-OrganizationinSmartEnvironment-Service-OrientedApproach
295
information is based on the usage of the common
RDF-compatible ontology models and common data
formats. It is a free to use, open source solution
available under the BSD license (Smart M3 at
Sourceforge, 2010). Communication between
software entities is developed via Smart Space
Access Protocol (SSAP).
The system has been partly implemented in the
Museum of Karl May Gymnasium History (The
Museum of Karl May Gymnasium History, 2013)
located in St. Petersburg Institute for Informatics
and Automation Russian Academy of Science
building.
The visitor downloads software for getting
intelligent museum visitors support. Installation of
this software takes few minutes depending on
operating system of mobile device (at the moment
only Maemo 5 OS is supported). When the visitor
runs the system for the first time the profile has to be
completed. This procedure takes not more than 10
minutes. The visitor can fill the profile or can use a
default profile. In case of default profile the system
can not propose preferred exhibitions to the visitor.
Response time of the Internet services depends
on the Internet connection speed in the museum,
number of people connected to the network, and
workload of the services. The average response time
does not exceed one second.
An example museum attending plan is presented
in Figure 5. It consists of five museums: The State
Hermitage, the Kunstkamera, the Museum of Karl
May Gymnasium History, St. Isaac Cathedral, the
Dostoevsky museum.
6 CONCLUSIONS
The paper presents a reference model and service-
oriented architecture for a multi-level self-
organization in a smart environment. The proposed
reference model enables a more efficient self-
organisation based on the “top-to-bottom”
configuration principle, which assumes conceptual
configuration followed by parametric configuration.
The service-oriented architecture makes it possible
to replace the organisation of the smart environment
with that of distributed service network. Application
of the approach is illustrated via a museum smart
environment case study.
The approach has some limitations. In particular, it
requires the corresponding services (transportation
& museums) to be available in the smart
environment. At the moment, some of such services
have been developed as prototypes and wrappers for
existing third-party services have been developed.
For development of a working application this issue
has to be kept in mind. Besides, the functioning of
the client’s applicatuion requires almost permanent
Internet connection. Today, mobile Internet usage in
foreign countries is quite expensive, however, we
expect the situation to improve in the nearest future.
ACKNOWLEDGEMENTS
The research presented is motivated by a joint
project between SPIIRAS and Nokia Research
Center. Some parts of the work have been sponsored
Figure 5: A sample of museum attending plan in a visitor mobile device in the center of St. Petersburg.
The State Hermitage The Kunstkamera
The Dostoevsky museum St. Isaac Cathedral The Museum of Karl May Gymnasium History
KMIS2013-InternationalConferenceonKnowledgeManagementandInformationSharing
296
by grants # 12-07-00298, # 12-07-00302, # 11-07-
00368, # 11-07-00045, and # 13-07-12095 of the
Russian Foundation for Basic Research, project #
213 of the research program “Intelligent information
technologies, mathematical modelling, system
analysis and automation” of the Russian Academy of
Sciences, and project 2.2 “Methodology
development for building group information and
recommendation systems” of the basic research
program “Intelligent information technologies,
system analysis and automation” of the
Nanotechnology and Information technology
Department of the Russian Academy of Sciences.
REFERENCES
Ambient Networks Phase 2. Integrated Design for
Context, Network and Policy Management,
Deliverable D10.-D1, 2006. Retrieved April 18, 2007,
from http://www.ambient-
networks.org/Files/deliverables/D10-D.1_PU.pdf.
Balandin, S., Moltchanov, D., Koucheryavy, Y., eds.
Smart Spaces and Next Generation Wired/Wireless
Networking, Springer, LNCS 5764, 2009.
De Mola, F., Quitadamo, R. Towards an Agent Model for
Future Autonomic Communications. In Proceedings
of the 7th WOA 2006 Workshop From Objects to
Agents. Retrieved March 25, 2013, from
http://sunsite.informatik.rwth-
aachen.de/Publications/CEUR-WS/Vol-204/P07.pdf.
Dey, A. K. Understanding and Using Context. In Personal
and Ubiquitous Computing J., 5(1), pp. 4-7, 2001.
Dey, A. K., Salber, D., Abowd, G. D. A Conceptual
Framework and a Toolkit for Supporting the Rapid
Prototyping of Context-Aware Applications. In
Context-Aware Computing. Moran et al. (eds), A
Special Triple Issue of Human-Computer Interaction,
16, pp. 229-241, 2001.
Honkola, J., Laine, H., Brown, R., Tyrkkö, O. Smart-M3
information sharing platform. In The 1st Int’l
Workshop on Semantic Interoperability for Smart
Spaces (SISS 2010) in conjunction with IEEE ISCC
2010, 2010.
Institute for Rexconfigurable Smart Components. Self-
Organizing Sensor Networks. Retrieved March 25,
2013, from http://www.eng.auburn.edu/users/lim/
sensit.html.
ITO Project S3. Scalable Self-Organizing Simulations,
2000. Retrieved March 25, 2013, from http://
www.dimacs.rutgers.edu/Projects/ Simulations/darpa/.
Johanesson, P., Andersson, B., Bergoltz, M., Weigand, H.
Enterprise Modelling for Value Based Service
Analysis. In The Practice of Enterprise Modelling
(Eds. by J. Stirna and A. Persson), First IFIP WG 8.1
Working Conference, PoEM 2008, Springer, LNBIP
15, 2008, pp. 153-167.
Raz, D., Juhola, A. T., Serrat-Fernandez, J., Galis, A. Fast
and Efficient Context-Aware Services, John Willey &
Sons, Ltd, 2006.
SENSEI. Integrating the physical with the digital world of
the network of the future. Retrieved March 25, 2013,
from http://www.sensei-project.eu/.
Serugendo, G. Di M., Gleizes, M.-P., Karageorgos, A.
Self-Organisation and Emergence in MAS: An
Overview. In Informatica, 30, 2006, pp. 45-54.
Smart-M3 at Sourceforge. Retrieved March 25, 2013,
from http://sourceforge.net/projects/smart-m3.
Smart-M3 at Wikipedia, Retrieved March 25, 2013, from
http://en. wikipedia.org/wiki/Smart-M3.
Smirnov, A., Kashevnik, A., Shilov, N., Oliver, I.,
Boldyrev, S., Balandin, S. Profile-Based Context
Aware Smart Spaces. In Proceedings of the XII
International symposium on problems of redundancy
in information and control systems, 2009 pp. 249 –
253.
Smirnov, A., Kashevnik, A., Shilov, N. Ontology
Matching in Context-Driven Collaborative
Recommending Systems. In Proc. of International
Conference on Knowledge Management and
Information Sharing (KMIS 2012), 2012, pp. 139-144.
SOCRATES. Self-Optimisation and self-ConfiguRATion
in wirelEss networkS. Retrieved March 25, 2013, from
http://www.fp7-socrates.org/.
Telenor. R&D Report, Project No TFPFAN, Program
Peer-to-peer computing. Retrieved April 18, 2007,
from http://www.telenor.com/rd/pub/rep03/
R_17_2003.pdf.
The Museum of Karl May Gymnasium History. Retrieved
March 25, 2013, from http://www.spiiras.nw.ru/
modules.php?name=Content&pa=shshowpa&pid=8
(in Russian).
Uschold, M., Grüninger, M. Ontologies: Principles,
methods and applications. In Knowledge Engineering
Review, 11(2), pp. 93-155, 1996.
Waterford Institute of Technology. Self Orchestrating
CommunIty ambiEnT IntelligEnce Spaces, 2013.
Retrieved March 25, 2013, from http://
cordis.europa.eu/earch/index.cfm?fuseaction=proj.doc
ument&PJ_RCN=11486081.
MultilevelSelf-OrganizationinSmartEnvironment-Service-OrientedApproach
297
