LINK™ PLATFORM

LINK™ COGNITIVE STREAMING ANALYTICS DELIVERS:

• Complex Event Processing: Ease of Integration to various data sources, to enable real-time detection and processing of network events and customer transactions.
• Real-time Analytics and Execution: Capability to act and decide based on real-time events as they relate to business metrics and objectives, creating a competitive advantage for businesses guaranteed only by customer loyalty.
• Real-time Predictive Modeling & Cognitive Loop: Learning and optimization in real-time based on customer feedback on offers
• Workbench & MPower Studio: Configurability of new capabilities, which allow users to create customized experiences in accordance to their brand values.

Introduction

LINK™ Cognitive Streaming Analytics uses innovative techniques in processing complex events and data streams in-memory. By combining machine learning capabilities and a high-performance workflow engine, LINK™ delivers continuously improving experiences tailored for each customer.
The cognitive capabilities of LINK™ bridge analytics and execution. Parameterized rules are configured to drive each business metric. LINK™’s machine learning capabilities determine how these parameters should be changed in order to achieve business targets. This process is known as the Cognitive Loop. LINK™ implements cognitive loops in high volumes (>100, 000 TPS) and velocity (milliseconds).
Cognitive loops can be applied beyond customer analytics and onto network analytics, product analytics, service analytics, and business process analytics. The result is a truly cognitive enterprise for any industry.

dtq-link-intro

LINK™ Architecture

LINK™ combines an industry proven multi-tier architecture and framework, in a single analytical platform. Still, each functional component is well defined. This modular design makes the platform highly scalable and highly available. This architecture also simplifies maintenance, deployments, and service deliveries.

Complex Event Processing

dtq-link-complexeventprocess LINK™ Complex Event Processing integrates with external data feeds. LINK™ provides real-time mediation and translation functions needed to convert any data format into an internal one, in preparation for analysis by other LINK™ components. The Complex Event Processing (CEP) can: – Integrate to telecommunications SS7 Circuit Switched (CS) and Packet Switched (PS) networks for real-time sessions – Integrate to both OSS and IT networks for file-based or protocol-based events – Retrieve network KPI’s from OSS or other 3rd party probes – Monitor key links in an IP network; by utilizing advanced deep packet capture and analysis technology, the CEP can work as a network tap. It uses GPGPU for fast processing.

Real-Time Analytics and Execution

dtq-link-realtime-analytics Once these continuous streams reach the real-time analytics tier, the Rules Engine applies the necessary analytical functions based on user-defined rules and criteria. All analytical processes are done in-memory. As events come in, “snapshots” of customer, network, product, and services profiles are kept in the In-Memory Database (IMDB). Analytical processes are performed against these snapshots. These processes can be a combination of data analytics and predictive analytics. The rules engine looks at the same snapshots and triggers the Workflow Engine when an action involving an external system needs to be initiated as a result of the analytics. These actions may include sending notifications, rewards, managing bandwidth, etc.

Real-Time Report

 

dtq-link-realtime-report

Snapshots from incoming events are collected and aggregated in the In-Memory Database based on report definitions. Reports are visualized through real-time dashboards. When Report KPI’s are defined, actions can be triggered when a report reaches a certain value or threshold.

Real-Time Predictive Modeling

dtq-link-realtime-modeling In cases where prediction is required, the Online Model Engine will be used to build, learn, and apply predictive models. The Online Model Engine uses streaming data to continuously update the model using a closed loop iterative execution. It improves the accuracy of the prediction, and in some cases changes the algorithm itself. This closed loop orchestration is the LINKTM learning system. Models created can be provided to the Recommendation Engine, which uses the aggregated model to give a recommendation or best offer to a customer.

Cognitive Loop Management

dtq-link-cognitiveloop The concept of Cognitive Loop is to improve the outcome of business rules. Parameterized business rules run through the Experience Database. Actual results of interactions based on the business rules are measured against defined Key Performance Indicators. By storing parameter-result pairs, the system can suggest the necessary changes in business rule parameters in order to meet business targets.

Tools: Workbench dtq-link-tools LINK™ Workbench is a web-based tool used for rules configuration and online system monitoring and management. This graphical user interface is a rich portal where users from marketing, sales, business intelligence, and operations teams can access LINK™’s functionalities to design and launch services, as well as monitor processes.

Tool: mPower Studio

 

dtq-link-mpower
MPower Studio is a software development kit (SDK) specifically designed to provide an easy, graphical, and drag-and-drop way to define a business process.. It is used for 3rd party system integrations during the platform deployment phase. MPower provides the capability to perform sizing, pre-system configuration, and other activities for engineers. It also stores known protocols and interfaces in a catalogue for re-use in future projects.