Entek Elektronik - Tibbo Aggregate IoT Anatomy

Tibbo Systems and AggreGate Platform

Transcript

1. Internet of Things Anatomy
2. Marke ng experts introduced the Internet of Things • There was no revolu on, just evolu on • ‘Things’ have been communica ng for quite a while (e.g. PLCs on a wire drawing line or network switches) • Monitoring and management systems have been exis ng for long, but again marke ng experts sent them to the ‘cloud’ • Cellular and satellite modems weren’t invented yesterday • In fact IoT is just a general name joining various markets, both B2B and B2C • Terminology evolu on: Intelligent Device Management => M2M => IoT - 2 -
3. Internet of Things comprises Devices (“things”) Data centers M2M concept assumes that devices interact with one another. They can do it: 1) Directly via network 2) Via network and central so ware in a data center (in the ‘cloud’) 3) Some mes both Networks - 3 -
4. Device Network Structure IP TCP, UDP SNMP, Telnet, BACnet, Modbus, SOAP, HTTP, MQTT… RS-232, RS-485, Ethernet, Wi-Fi, USB, CAN, Bluetooth Z-Wave, GPRS/3G/LTE… PPP, ATM, SLIP… NetBIOS, PPTP, RPC… SSL, TLS... - 4 - OSI Network Model
5. Device Types The difference is in management so ware tasks. Example: GPS trackers for a dog and a bus are similar in terms of hardware, but they have absolutely different could services and dashboards. Consumer Industrial - 5 -
6. Device Logical Structure Variables (se ngs, proper es): ability to read and write Such device structure is described in full or par ally by any known communica on protocol. Func ons (methods, opera ons): ability to call and transmit input data while receiving output data Events (no fica ons): ability to subscribe and retrieve instances asynchronously Metadata (descrip ons of available variables, func ons and events) - 6 -
7. Internet of Things Pla orm • IoT pla orm is just a regular server so ware • It plays a role of run me environment (applica on server) for IoT applica ons designed for the end user - 7 - • Only a few applica ons are wri en from scratch • IoT pla orms are o en deployed in rented commercial data centers, or in data centers belonging to large IoT device operators
9. Types of IoT Pla orms • Infrastructure pla orms provide data storage and collec on as well as API/SDK for implemen ng processing, visualiza on and integra on methods (IoT applica on development) via programming • Full cycle pla orms solve all tasks using visual constructors, with the only necessity for programming when wri ng communica on modules and complex mathema cs/logic - 9 -
10. Communica on with Devices • Any IT (SNMP, Telnet, WMI...), automa on (Modbus, BACnet, OPC…), IoT (MQTT, XMPP, AMQP…) and universal (HTTP/REST, SOAP, FTP…) protocols are used • Very few basic opera ons: reading and wri ng se ngs, execu ng opera ons, receiving events (including no fica ons on change in values) - 10 -
11. Data Normaliza on - 11 - Normaliza on is conversion to a unified standard form. It’s usually performed in two steps: • Abstrac on from protocol (conversion to universal data types) • Abstrac on from device type/make/version (applica on of device models)
12. Data Storage - 12 - What we store: • Server-side configura on and tools • Last device configura on snapshots (in case of unavailability) • Se ng change history (for devices and server-side tools) • Event history (the same as above) Where we store: • Rela onal database (slow and inefficient) • NoSQL database (оптимально) • Specialized databases (e.g. RRD for me series aggrega on – has its own pros and cons) RDBMS RRD (Sta s cs) NoSQL (Big Data)
13. Data Processing - 13 - • Completely standalone • Delayed group configura on and opera on execu on • Operator no fica ons upon important events and states (emails, SMS) • Dynamic models with own life cycle • Machine-readable knowledge base for taking decisions • Mul ple tools (root cause analysis, scheduler, domain- specific languages – examples: AggreGate and IEC languages)
16. Why not to write everything yourself? - 16 - • A prototype will be ready quickly • You will spend years implemen ng a scalable system suppor ng failover clustering, distributed collec on and storage architecture, etc. • A bicycle will be invented in about 5 years • There’ll be fixed expenses to support the real product state • It looks even more unnatural for system integrators, engineering companies and MSPs
17. Tibbo Systems and AggreGate Pla orm - 17 - • Tibbo Systems: Russian so ware developer working worldwide • AggreGate Pla orm: so ware “brick set” for building IoT device monitoring and management systems • 14-years’ investments into “brick” development • Hundreds of large installa ons in various countries • 10+ ver cal market solu ons, including IT infrastructure management and SCADA systems
18. Cases and References - 18 - • Monitoring and managing telecom tower power supply (Flexenclosure, Sweden) • Monitoring mission-cri cal uninterrup ble power supply units (Unified Energy Corpora on, Russia) • Narrow-band radio sta on monitoring system (DCI Tech, Canada) • Comprehensive monitoring of a mul -server telecom operator network (An-net, Russia) • Monitoring of engineer construc ons (Insight, Russia) • Centralized fountain management (Sharel, Israel) • Roadheading equipment monitoring (Ilma, Russia)
19. Cases and References - 19 - • Building automa on of the Electoral Commission of Namibia • Data acquisi on from industrial alcohol breath tes ng devices (Intoximeters, the US) • Forkli fleet management and monitoring (Keytroller, the US) • Monitoring McAuto queue length and POS equipment (McDonald’s, the US) • Centralized monitoring, control and provisioning of Android-based vending machines (Minibar Systems, the US) • Cloud-based Time and A endance system (RCPOnline, Poland) • Monitoring of a distributed IP-based emergency no fica on speakers network (Emergencies Ministry of Russia)