Periodic Reporting for period 2 - SONNET (Self-OrganizatioN towards reduced cost and eNergy per bit for future Emerging radio Technologies)
Période du rapport: 2019-01-01 au 2022-06-30
SONNET considered a complex context-aware heterogeneous network that is slowly becoming a 5G reality, and merged the notion of SO (self-organising) with radio connectivity protocols such as network sharing and CoMP (Coordinated Multipoint) to enhance the SON (SO Networks) coverage within the mobile network in contrast to today´s network. We investigated the notion of SO CoMP and SO network sharing, as key technologies to reduce cost and energy per bit in legacy and future emerging mobile technologies.
Project Objectives:
Objective 1: Context-Aware Architectures for RAN (Radio Access Networks);
Objective 2: SO Network Sharing for Wireless Multi-Hop Heterogeneous Networks;
Objective 3: SO-CoMP for Cost and Energy per Bit Reduction;
Objective 4: Experimentation of SONNET Solutions Under Real Practical Conditions
Impact on Society:
In a mobile ecosystem that currently hosts a plethora of mobile operators, including virtual counterparts, competition is as strong as ever, raising the bar further on the quality of any potential mobile product entering the market. The introduction of new mobile services/products will come at a price: higher energy consumption and higher costs for maintaining the network. Mobile operators have to solve those two contentious issues in order to successfully continue offering competitive services and benefit the connected society at large. SONNET aims to address these issues by providing enhanced mobile services with lower costs and energy consumption through employing context-aware SON.
Project Objectives:
Objective 1: Context-Aware Architectures for RAN (Radio Access Networks);
Objective 2: SO Network Sharing for Wireless Multi-Hop Heterogeneous Networks;
Objective 3: SO-CoMP for Cost and Energy per Bit Reduction;
Objective 4: Experimentation of SONNET Solutions Under Real Practical Conditions
Impact on Society:
In a mobile ecosystem that currently hosts a plethora of mobile operators, including virtual counterparts, competition is as strong as ever, raising the bar further on the quality of any potential mobile product entering the market. The introduction of new mobile services/products will come at a price: higher energy consumption and higher costs for maintaining the network. Mobile operators have to solve those two contentious issues in order to successfully continue offering competitive services and benefit the connected society at large. SONNET aims to address these issues by providing enhanced mobile services with lower costs and energy consumption through employing context-aware SON.
• Identified SONNET scenarios and use-cases, system architecture, requirements and performance indicators
• Identified a Context-Aware Architecture for Self-Organization and context-based scheduling applications (figure 1)
• Developed a context-based battery priority scheduling algorithm for limited battery powered devices utilizing high data rate demanding applications.
• Proposed a Multi-domain SDN MEC architecture (figure 2) for resource provisioning. In particular, a deadlock aware algorithm for scheduling resources for Industrial IoT (IIoT) devices onto a MEC platform which incorporates the banker’s resource-request algorithm is proposed.
• Design and development of an energy-aware routing algorithm for SDN based Networks
We proposed a novel Temporal Resource-aware Routing Algorithm for SDN (STR-RA). Performance of the algorithm is verified using a GNS3 based implementation with an Opendaylight controller. The simulation results were compared against legacy RIP (Routing Information Protocols) and OSPF(Open Shortest Path First) protocols, where a gain of up to 6 orders of magnitude was observed in E2E (End-to-End) delay reduction.
• Design of SO-Network Sharing for heterogeneous networks (figure 3)
A mobile networking scenario was proposed that exploit the context-aware architecture paradigm, and SON module to gather context information about the user Signal-to-Noise and Interference in order to maximise the coverage-capacity of the network under a network sharing regime, as well to ensure energy efficient connectivity.
• Design of novel SO-CoMP transmission approaches for beyond 5G networks
SONNET implemented a hybrid approach based on centralized and distributed self-organized small cells that achieves low signalling overhead through direct small cell communication. The SON approach is shaped by two different optimization approaches, that include machine learning (ML) and game theory.
• An integrated SONNET testbed was developed, that is able to emulate interconnected heterogeneous infrastructures within a cloud based framework; this involved the integration of an in-house network (LSBU) and system level simulator (GS) with a commercial ray tracer tool (SIG). In particular, the SONNET deployment entertains a novel SO-KDN (Self-Organised Knowledge-Defined Networking) model which provides beyond-5G self-organization (Self-Optimization, Self-Configuration, Self-Healing and Self-Learning) on a Knowledge-Defined Network. A comprehensive set of simulation results were given too demonstrate how SO and machine learning in synergy can optimise network performance for the routing-as-a-service use-case.
The dissemination plan led to 32 published works (20 journals, 12 conference papers), 3 international workshops, and several key note talks and tutorials. The exploitation plan led to influencing the standards community through publication in the IEEE Communications Standards Magazine (SDN-Sim: Integrating a System-Level Simulator with a Software Defined Network; 4(1), March 2020). The scientifc outcomes also provided the basis for tutorials (for e.g. CollaborateCom 2019, AI-as-Service), and content for laboratory coursework (new experimentation problem-solving) that was included in the MSc program at the Uni. of Patras (Greece) and London South bank University (UK).
• Identified a Context-Aware Architecture for Self-Organization and context-based scheduling applications (figure 1)
• Developed a context-based battery priority scheduling algorithm for limited battery powered devices utilizing high data rate demanding applications.
• Proposed a Multi-domain SDN MEC architecture (figure 2) for resource provisioning. In particular, a deadlock aware algorithm for scheduling resources for Industrial IoT (IIoT) devices onto a MEC platform which incorporates the banker’s resource-request algorithm is proposed.
• Design and development of an energy-aware routing algorithm for SDN based Networks
We proposed a novel Temporal Resource-aware Routing Algorithm for SDN (STR-RA). Performance of the algorithm is verified using a GNS3 based implementation with an Opendaylight controller. The simulation results were compared against legacy RIP (Routing Information Protocols) and OSPF(Open Shortest Path First) protocols, where a gain of up to 6 orders of magnitude was observed in E2E (End-to-End) delay reduction.
• Design of SO-Network Sharing for heterogeneous networks (figure 3)
A mobile networking scenario was proposed that exploit the context-aware architecture paradigm, and SON module to gather context information about the user Signal-to-Noise and Interference in order to maximise the coverage-capacity of the network under a network sharing regime, as well to ensure energy efficient connectivity.
• Design of novel SO-CoMP transmission approaches for beyond 5G networks
SONNET implemented a hybrid approach based on centralized and distributed self-organized small cells that achieves low signalling overhead through direct small cell communication. The SON approach is shaped by two different optimization approaches, that include machine learning (ML) and game theory.
• An integrated SONNET testbed was developed, that is able to emulate interconnected heterogeneous infrastructures within a cloud based framework; this involved the integration of an in-house network (LSBU) and system level simulator (GS) with a commercial ray tracer tool (SIG). In particular, the SONNET deployment entertains a novel SO-KDN (Self-Organised Knowledge-Defined Networking) model which provides beyond-5G self-organization (Self-Optimization, Self-Configuration, Self-Healing and Self-Learning) on a Knowledge-Defined Network. A comprehensive set of simulation results were given too demonstrate how SO and machine learning in synergy can optimise network performance for the routing-as-a-service use-case.
The dissemination plan led to 32 published works (20 journals, 12 conference papers), 3 international workshops, and several key note talks and tutorials. The exploitation plan led to influencing the standards community through publication in the IEEE Communications Standards Magazine (SDN-Sim: Integrating a System-Level Simulator with a Software Defined Network; 4(1), March 2020). The scientifc outcomes also provided the basis for tutorials (for e.g. CollaborateCom 2019, AI-as-Service), and content for laboratory coursework (new experimentation problem-solving) that was included in the MSc program at the Uni. of Patras (Greece) and London South bank University (UK).
Context-aware Energy Efficient Packet Scheduling for 5G mobile networks:
SONNET investigated a context-aware packet scheduling (CAS) algorithm which reduces the energy consumption by considering energy related parameters within the the context information.
SO-Network Sharing:
SONNET went beyond by merging SO with network sharing to provide an integrated and adaptive solution to optimise network coverage-capacity. The situation of employing network sharing is assumed through operators having a common service-level agreement.
IP Routing:
SONNET proposed the design and development of an energy-aware routing algorithm. In particular, it proposes a resource-aware routing algorithm for SDN, which monitors the resource utilization of network devices (nodes) and channels (links) using a push agent to fetch the topology and flow table information from the controller. Using link queue modelling and stochastic network calculus, it guarantees a route that avoids busy nodes and exploits underutilsed links. Results show the validity of the algorithm.
MEC and Resource Provisioning:
SONNET proposed a novel resource provisioning algorithm for deadlock avoidance in multi-access edge computing in the context of IIoT. A deadlock aware algorithm is proposed for scheduling resources based on the banker’s resource-request algorithm. The Banker’s algorithm works by simulating and using specified resources to predetermine deadlock conditions for all pending activities and deciding if allocation should be allowed to continue. The proposed algorithm is only favorable if implemented using Software Defined Networking (SDN) to reduce the communication overhead that would be generated by the resource-request algorithm.
SO-CoMP Transmission:
SONNET went beyond SOTA by combining SON with CoMP resulting in what we refer to as a novel hybrid mode of self-organised CoMP that exploits the benefits of both the centralized and distributed interference management (IM) approaches.
Product Innovation:
We go beyond the current capabilities of each individual partner tool to provide an integrated experimental platform (figure 4) that collectively are able to model the future emerging 5G mobile networking paradigm as a virtual pool of networking resources. Moreover, this platform was used to validate the innovations on SO-Network Sharing and SO-CoMP through the use of context-aware information.
SONNET investigated a context-aware packet scheduling (CAS) algorithm which reduces the energy consumption by considering energy related parameters within the the context information.
SO-Network Sharing:
SONNET went beyond by merging SO with network sharing to provide an integrated and adaptive solution to optimise network coverage-capacity. The situation of employing network sharing is assumed through operators having a common service-level agreement.
IP Routing:
SONNET proposed the design and development of an energy-aware routing algorithm. In particular, it proposes a resource-aware routing algorithm for SDN, which monitors the resource utilization of network devices (nodes) and channels (links) using a push agent to fetch the topology and flow table information from the controller. Using link queue modelling and stochastic network calculus, it guarantees a route that avoids busy nodes and exploits underutilsed links. Results show the validity of the algorithm.
MEC and Resource Provisioning:
SONNET proposed a novel resource provisioning algorithm for deadlock avoidance in multi-access edge computing in the context of IIoT. A deadlock aware algorithm is proposed for scheduling resources based on the banker’s resource-request algorithm. The Banker’s algorithm works by simulating and using specified resources to predetermine deadlock conditions for all pending activities and deciding if allocation should be allowed to continue. The proposed algorithm is only favorable if implemented using Software Defined Networking (SDN) to reduce the communication overhead that would be generated by the resource-request algorithm.
SO-CoMP Transmission:
SONNET went beyond SOTA by combining SON with CoMP resulting in what we refer to as a novel hybrid mode of self-organised CoMP that exploits the benefits of both the centralized and distributed interference management (IM) approaches.
Product Innovation:
We go beyond the current capabilities of each individual partner tool to provide an integrated experimental platform (figure 4) that collectively are able to model the future emerging 5G mobile networking paradigm as a virtual pool of networking resources. Moreover, this platform was used to validate the innovations on SO-Network Sharing and SO-CoMP through the use of context-aware information.