[vc_row el_class="section"][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_single_image image="42048" alignment="center" css=".vc_custom_1674626162140{margin-bottom: 0px !important;}"][vc_separator][vc_custom_heading text="Orange: A Digital Twin Platform to Make Smart Cities Even Smarter" font_container="tag:h1|text_align:left" use_theme_fonts="yes"][vc_custom_heading text="With ArangoDB at its core, Orange has created 51 million digital twins and counting" font_container="tag:h4|text_align:left" use_theme_fonts="yes"][vc_empty_space][vc_column_text]Results:

  • Creation of Thing in the Future digital twin platform, providing a system vision of the physical world
  • Simplified architecture due to ArangoDB’s support for graph, key/value, and document data models
  • A foundation to help develop services that can power smart cities, industrial environments, smart agriculture, and more

[/vc_column_text][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_custom_heading text="The Scenario: Going beyond the Internet of Things" use_theme_fonts="yes"][vc_column_text]Orange is one of the world’s leading telecommunications operators, with a total customer base of 271 million customers worldwide in 2022, sales of 42 billion dollars, and 140,000 employees worldwide. Orange is also a leading global IT and telecommunication service provider to multinational companies under the brand Orange Business Services.

In 2019, Orange’s research group embarked on a project to create a platform that took the capabilities of classical Internet of Things (IoT) platforms a step further. Shares Thomas Hassan, Researcher at Orange, “The main objective that we wanted to achieve was being as universal as possible – breaking the silos of data and domains to be able to build new use cases and new applications that can span logistics, industries, smart cities, healthcare, and more. Applications that are truly cross-vertical.”

To accomplish these objectives, Orange set out to build a digital twin platform that describes the physical world through connected sensors and objects, along with their interactions. Such a platform could benefit the development of services that power smart cities, smart buildings, industrial environments, or smart agriculture. For example, entities modeled in a digital twin of a smart city can include water and electricity, waste management and recycling, public transportation, housing, roads, parking lots, trees, and traffic lights – all of which are owned and managed by different actors (municipality, region, state, or private sector).[/vc_column_text][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_custom_heading text="The Requirements: A graph database to power a digital twin platform" use_theme_fonts="yes"][vc_column_text]To accommodate such a diverse data set and map the relationships between them, Orange knew it needed a graph database to structure the information as a digital twin. Explains Hassan, “We see graph as a very versatile way to structure information. The key point is that we can capture system of systems composition and model very complex systems. By bringing together and adding information to build up the graph incrementally, it can become richer and richer with every added link.”

Orange’s requirements for the graph database it wanted to use included:

  • Ability to accommodate diverse data
  • Scalability and geo-distribution capabilities
  • Support for ACID transaction processing
  • Powerful indexing, including full-text and geo

[/vc_column_text][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_custom_heading text="Why ArangoDB: Diverse data model support complete with GeoJSON and ACID" use_theme_fonts="yes"][vc_column_text]After running into scalability roadblocks with OrientDB, Orange turned to the ArangoDB graph database. They were pleased to find ArangoDB offered the following features to support their use case:

Support for diverse data models: Although ArangoDB is a highly-scalable graph database, it also offers the capability to store data as key/value pairs and documents – all within the same core, united by a single query language. This allowed Orange to minimize complexity without sacrificing flexibility.[/vc_column_text][vc_custom_heading text="“ArangoDB’s support for a diverse, broad, and versatile data model is a key advantage to simplifying our complex architecture.”

- Thomas Hassan, Orange" font_container="tag:h2|text_align:left|color:%23a3b34f" use_theme_fonts="yes" el_class="course-header"][vc_empty_space][vc_column_text]Full-text and geo-spatial indexing: ArangoDB features a Google S2-based geo-spatial index, supported for a subset of the GeoJSON geometry types and simple latitude/longitude pairs. It also includes ArangoSearch, a built-in, full-text search engine that can index nested fields from multiple collections, rank query results by relevance, and more.

Transaction and ACID support: ArangoDB offers three types of transactions, all of which are ACID: AQL queries, Stream, and JavaScript.[/vc_column_text][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_custom_heading text="The Implementation: Digital twins, semantic modeling, advanced search, and more" use_theme_fonts="yes"][vc_column_text]With ArangoDB at its core, Orange built Thing in the Future, better known as Thing’in. Thing’in is a multi-sided, open digital twin platform composed of physical things and entities, and the relationships between them. It allows IoT application developers to connect to and interact with the front end, and data providers to do so at its back end. As of 2022, Thing’in had 51 million devices and 44 million relationships.[/vc_column_text][vc_single_image image="42158" img_size="large" add_caption="yes" alignment="center"][vc_column_text]The core functionalities of Thing’in are:

Storage of the digital twin, or what Orange calls an ‘avatar’, and its set of properties and functionalities. Each ArangoDB node represents an avatar, and the graph’s edges connect the relationships between avatars. The relationships between the avatars allow Thing’in to create a systemic vision of the physical world. Stored as key/value attributes, properties may be associated with avatars and relationships. 

Semantic modeling. Through the use of shared vocabularies, Thing’in can put semantic definitions on the digital twins and their relationships.

Advanced search. Powered by graph traversals, users of the Thing’in platform can search through avatar properties, geolocations, and semantics.

Clustering allows Orange to provide a systems of systems view, making clusters of digital twins together to help manage large fleets of IoT devices or twins.

Access modalities to help guides users on how to access object data and make actions to it.

Historization to store all the events and go back in the past to analyze what happened.

Orange currently has two deployed platforms of Thing’in: one development platform and one pre-production platform. The development platform consists of three node clusters, 218 users, and 8 million nodes, while the pre-production platform is a single-node deployment with 264 users and 52 million nodes.  Thing’in is currently hosted on Orange’s cloud infrastructure; it will be also deployed soon on-premises and on other cloud servers managed by third parties.  Here’s a diagram of Thing’in’s global architecture:[/vc_column_text][vc_single_image image="42160" img_size="417 x 500" add_caption="yes" alignment="center"][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_custom_heading text="The Results: A graph-centric Web of Things Platform" use_theme_fonts="yes"][vc_column_text]With ArangoDB at the core of Thing’in, Orange was able to build a digital twin platform that provides the following benefits:

  • A digital model of physical assets to enable a better understanding of a domain-specific process or activity;
  • Optimize processes or how a physical asset works, as well as predict what will happen in the future, by utilizing simulation use cases;
  • Reduce costs of training employees when combined with technologies such as augmented reality; and
  • Set up preventive and predictive maintenance plans, as well as capacity planning.

[/vc_column_text][/vc_column][vc_column width="1/12"][/vc_column][/vc_row][vc_row][vc_column width="1/12"][/vc_column][vc_column width="10/12"][vc_column_text]For more details, you can watch Thomas' presentation from ArangoDB Summit 2022:[/vc_column_text][vc_video link="https://arangodb-1.wistia.com/medias/agpwrul7u4"][/vc_column][vc_column width="1/12"][/vc_column][/vc_row]