In the world in the near future, things and applications will be connected across companies and industries. What appearance will the ICT platform to support the new world have? This article introduces the latest trends of the smart coordination technology in the ICT/network resource service.

The World Where Various Things and Applications Will Be Connected

Currently, service providers are developing their own things (devices) and applications in order to launch services on their own. We think that the time will come to be able to provide various services to use IoT (Internet of things) and AI(artificial intelligence) by linking the things and the applications provided by a variety of enterprises and industries. In such time, we think, a new ICT platform will be wanted to easily use necessary things and applications as many as they are needed and only when they are needed.

NTT, for the realization of such world, has been working on the researches and developments of the smart coordination technology for ICT/network resource services. In the direction of our researches and developments, first of all, not only the overall constructing ability involved in the existing smart coordination functions but also the ingenuities (architecture) for the smart coordination function to be able to flexibly handle the diversification of services and users and the increase of the data to be processed will be the technical features as well as the high-level assurance functions to support them.
Furthermore, we think some intelligent functions to use the AI technology will be needed for providing further high-level services including providing the optimal combinations to match the requirements of service providers and users and providing the optimal service-implementing environments for the use conditions and environments.
Here, we show you the future image of the ICT platform envisioned by NTT together with its technical features (Fig. 1).

〈Flexible〉 - The Operation Model to Promote Flexibility and Automation -

Service providers have been taking it upon themselves to combine other companies' open resources/functions in order to build new services (smart coordination services) in response to the current trends of cloud-first. In such trends, it has become essential to speedily provide the combinations of various resources and functions. For this purpose, automation of operation will also be necessary for on-demand provision of resources/functions.

Even though automation technologies are already widespread in the cloud service operation of today, network service operation is not yet sufficiently automated partly under the influence of existing systems in the networking services and the like. However, in recent years, SDN (Software Defined Network) is beginning to be introduced as the idea for virtualization and, thus, the device environment for the realization of providing on-demand services is beginning to be organized. Under these circumstances, we have been gradually advancing our examinations for the enhancement of the flexibility of operation as well as its automation.

●Catalog-driven orchestration and operation model

For our first step, we have discussed, as a use case, "catalog-driven type orchestration technology," which is the automation technology to be able to flexibly handle the combination of various resources and functions by simplifying, in the style of catalogs, the pre-process settings necessary for the comprehensive construction of the external cloud services and mobile lines whose service application functions are open as APIs ^[1]. Part of the achievement was used also in the demonstration experiment conducted in Las Vegas in 2018, contributing to make efficient environment constructions ^[2].

As the next step, we have started examinations for the operation process automation including customer and contract management in addition to service management. For this purpose, we think, it is necessary to establish an operation model to clearly show the relation of the API for formulating the operation processes and for circulating the information needed to execute necessary functions and operations as well as the information among functions.
Also, TM Forum, having been examining the operations by telecommunications operators, on the basis of the trend in the use of APIs, is advancing the reviews of the past operation processes and data models as well as the definitions of functions together with making linkage with APIs ^[3]. We, noticing such move and other trends of standardization and external trends, have been advancing the formulation of operation models.

●API Design Support Technology

The public APIs are frequently of the specifications unique to service providers in terms of the rules for describing API requests and responses, the methods of authentication, and the like. As the specifications are different as mentioned above, the parties to develop services have difficulty in using APIs. In addition, as for the parties to provide APIs, they may experience the risk of the occurrence of some unnecessary discussion at the time of designing if there are not rules defined for publication and/or description.
We, also in consideration to the trends of the standards in our industry, have been advancing the definition of the rules for API description. In addition, we have been examining the technology for giving support to the conformity to the defined rules and the technology for determining whether APIs conform to such rules at the time of designing API specifications. For the support to such conformity, we have examined a method that uses the Swagger specifications ^[4], the format widely used for describing API specifications and that outputs the templates conforming to the rules for description only by giving answers to the simple questions with respect to API publication.

●Technology to Enhance the Assurance Function

The assurance operation today is based on a variety of information necessary for assurance operation including service structures, service specifications, user information, on-market technologies, combination-source APIs, trouble notices, test calls, and SLA (Service Level Agreement), and the assurance operation service providers with a wide range of knowledge and know-how require a large number of occasions for operation to make decisions and to control them.
If this is a smart coordination service combining two or more services, the assurance operation service provider will be under further burden. This is why we have been examining the following technology to establish the service assurance enhancing technology that will make it easier to be able to follow new services not only by using the automation of the operation on the side of the assurance operation service provider of a smart coordination service but also to aim to establish the service assurance enhancement technology (Fig. 2).

① The autonomous cooperative type architecture for assurance operation
② The assurance information collecting type based on the stationed monitoring agent type

⇒① The assurance type in which the assurance operation is divided into parts and individual assurance parts make autonomous judgment and use messaging to share information and cooperate with each other without intervention from human labor.
The assurance process is implemented by flexibly combining the assurance function parts to make efficient the assurance operation to follow smart coordination services

⇒② The type in which the programs to be able to collect the flow information of the service user and other detailed information (monitoring agent) are stationed flexibly in appropriate places.
The type of the assurance information to be processed and the monitoring places are expanded so that, even in the cases where it is not possible to identify the places of troubles based only on the information acquired from the APIs provided in an existing network, cloud service, and/or the like, it will be enabled to increase the information available for analyses and/or judgments and thus, even in the smart coordination service that combines two or more services from different service providers, the end-to-end monitoring wanted from the viewpoint of service users will be realized

〈Intelligent〉 - "Tacit Computing," the Endeavors for Providing Advanced Services -

Tacit Computing is a generic name for IoT and AI service platform technologies whose research and development are conducted by NTT Network Service Systems Laboratories. To construct a service by combining necessary on-demand things and applications, it is necessary to grasp the status of the devices and computers and the requirements of the service and to appropriately coordinate them. To cope with such tasks, we have been working on the following three elemental technologies to construct service systems by appropriately combining "Behavior Automatic Analysis Technology" to automatically distinguish the types and models of the devices connected to the network together with other devices and/or software programs (Fig. 3).

●Relation-Location Matching

The technology to appropriately select devices, software programs, and networks for constructing a service form two or more possible options. To assure the quality of service, it is needed not to cause excessive burden on networks or computers even if two or more services use the same device and/or software at the same time. This technology, based on the types and the use statuses of the devices connected to the network, the status of the network, the requirements and conditions of the service, and the like, determines the smart coordination (relation) of devices and/or software programs and the executing environment (location) of software. The processes and data common to two or more services are integrated to make efficient the overall process and, furthermore, the location of software will be appropriately selected based on the network bandwidth, communication delay, and so forth.
For instance, when a suspicious person detection service and a lost child finding service are executed at the same time, the human image detection process, which is common to both service, is extracted and will be integrated and executed on the computer close to the camera in the searched area.

●AI Resonance

The technology that automatically coordinates multiple devices and software programs upon on-demand request. To build a service, it is necessary to choose appropriate settings that suit the type of device, its location, and software. In conventional services, appropriate settings have been made through the trial and error made by humans. However, in the world we are looking for, the combinations of the devices and software programs to construct a service are huge and are varying as time goes by; therefore, it is already difficult for humans to make such settings. In this technology, the result of the behaviors of devices is evaluated automatically in real time to learn appropriate settings.
For example, in a lost child finding service, the lighting to make clear the video images is automatically selected in the network to make adjustment to an appropriate lighting level. As appropriate settings are automatically made while the service is executed, it is not necessary for humans to make designing or adjustment; thus, it becomes possible to provide an on-demand service at an advanced level.

●Environment Adaptive Code Generation Technology

The technology that automatically optimizes programs depending on the hardware to execute them. Some computers in a network may have special hardware such as GPU or FPGA (Field Programmable Gate Array). It is wanted to be able to exercise some high-level performance by selecting specific computers as the execution environments of software. In this technology, the source code of a program will be automatically converted into the form that is suitable to its execution environment. As an example, we confirmed that the performance should be enhanced by about four times by automatically extracting and converting the processes to be off-loaded into the GPU among the image analysis programs. In coming years, we aim to adapt this to further various environments including a variety of IoT devices and quantum computers, which will be our new hardware.

Future Development

The scalable architecture and the technology to assure and advance it are the technologies to add value to the existing orchestrator. In coming years, we are going to advance our research and development on such technology for scalable application and are planning to also advance the enhancement of the technology by proactively using the AI technology for the service assurance technology per se ^[5]. Moreover, as for Tacit Computing, we are going to advance the establishment of the elemental technologies on optimal resource selection and setting automation and, in addition, aim to realize <Open IoT> for free and mutual smart coordination among devices and application programs through our research and development on adapting various IoT resources and services.

*This essay comprises from NTT Technical Review: Technology for Smart Coordination of ICT/Network Resources and Services, July 2019, Vol. 17, No. 7.