4. IoT aware intelligent IP network’s behavior prediction

a. Description:

Modern IP networks are already challenged by IoT, which requires high availability, adequate IP SLA parameters (delay, throughput, packet loss, jitter etc.) and extensive IP network scalability / flexibility. SmartNIM (Network Infrastructure Monitoring) is a solution for 24/7 active monitoring of the vital network infrastructure. It consists of active SmartNIM probes and central data acquisition system for reporting visualization and alerting. SmartNIM probes are placed on the vital network endpoints / locations. Existing SmartNIM probes are running on x64 Docker probe, ARM SBC (for example on Raspberry PI v3). For communication between probes and central data acquisition system MQTT protocol is used. Measurement results from probes with probe related data are stored in Elasticsearch. Each SmartNIM document in Elasticsearch consists of: timestamp, geoLocation, testType, testId, probeId and test result values (depending on the testType).

b. Outcomes:

Implementation of the Machine learning principles as a basis for IoT aware intelligent IP network’s behaviour prediction. Network quality parameters should be correlated with public accessible data source (e.g. ARSO, GIS, Scalar etc.). Visualisation should be done through creative graphic tools (for example: bubble chart, radar graph etc.).

SmartNIM Introduction:

SmartNIM is an end-to-end solution enabling SLA monitoring and reporting on various network related KPI’s.

Many organizations, where ICT infrastructure and services provided are required to be resilient, highly available and secure, require monitoring of SLA parameters to ensure required levels of network services operations.

SmartNIM probes are placed on the vital points in the network, from where critical network infrastructure is actively monitored. Smart NIM probes can be hardware SBC’s platforms (such as widely available platforms as Raspberry Pi or BananaPro) or can be deployed integrated with the networking equipment. SmartNIM probes can also be deployed on any orchestration platform (MS Hyper-V, VMware, OpenStack, Docker containers, …).

Results are reported to the central SmartSNO server where network infrastructure monitoring records are stored and threshold checking and alerting is done. SmartSNO enables visualization and reporting on the network infrastructure monitoring records. Service Level Agreements (SLA) reports are also supported by SmartSNO.

Alternatively, results can be placed on any other storage/reporting platform either SQL or NoSQL. Logstash / Elasticsearch / Kibana is already integrated with the SmartNIM probes. SmartNIM Probe measurement results are processed by the Logstash and placed into the Elasticsearch. Kibana is used as a frontend for the summarized tabular / graph data representation.

There are numerous tests that can be performed using SmartNIM:

  • IP connectivity monitoring
    • Availability of the vital IP network infrastructure (local gateway router, wireless controller, …)
    • Quality of the IP network “SLA – KPI” parameter’s (packet loss, delay and jitter)
    • Bandwidth measurements
      • ICMP bandwidth throughput measurement’s (bandwidth throughput indicator)
      • UDP limited or TCP limited / unlimited bandwidth measurements
    • Network resources monitoring
      • IP network addresses engaged (periodical IP networks scanning)
      • Seen MAC addresses / vendors scanning (local network only)
      • Periodical WLAN network scanning
      • WLAN SID access points monitoring (depends on the platform)
    • Monitoring of the vital network related services
      • DHCP service
      • DNS service
      • MS AD service *
      • Radius service *
      • NTP service *
      • MqTT service
    • Vital business related services
      • MS SQL service
      • e-mail service *
      • HTTP/HTTPS web services *
      • file server services *
      • VoIP service *
    • Security related tests
      • port scanning *
      • IP address scanning


* in the roadmap

SmartNIM Benefits

  • Instant setup (no probes on the production servers)
  • Highly customizable (new service types can be easily added)

SmartNIM Use Cases

  • Service operators
    • Monitoring of the priority customers’ infrastructure
    • SLA collecting and reporting
  • Enterprise:
    • Monitoring remote locations (is the remote WLAN running, DNS, DHCP ok, unregistered IP address detected, …)
    • Supervising TK services provided by service providers

Large MqTT infrastructure monitoring

SmartNIM HackElect Infrastructure

One Intel x64 based SmartNIM probe is used as Central SmartNIM probe. On the same hardware the VPN server, MqTT broker and MqTT to AWS Elastic gateway are installed.

First SmartNIM Raspberry Pi probe (peerLocation = SC) is installed and connected to the wired internet – aim of this probe is to be an “infrastructure reference probe”.

Four SmartNIM Raspberry Pi probes are installed and connected to the mobile LTE 4G network. All “mobile” probes are using same mobile operator and are geographically distributed. VPN client is configured on “remote” SmartNIM probes. All tests are initiated by SmartNIM Central Probe.

All communication between probes is tunneled trough private VPN.