Across Methodology
ACROSS is poised to become a sui generis E2E service deployment and management platform for next-generation
networks and services, aiming at unprecedented levels of automation, performance, scalability, and energy efficiency. To accomplish this challenging mission, ACROSS capitalises on eight important axes.
Scalable end-to-end service orchestration platform
built as a highly-distributed grid of domain-level orchestrators, spread across multiple geo-dispersed and potentially heterogeneous edge environments (e.g., for industries, agriculture, logistics), overseen by a cloud-managed multi-domain orchestrator, which ensures end-to-end principles over the ever-expanding edge-to-core cloud continuum.
Standardised communication interfaces
foster separation of concerns between the multi-domain end-to-end orchestrator and the various distributed domain-level orchestrator instances through the ACROSS integration fabric. This fabric ensures compliance with ongoing standardisation efforts, such as the ETSI Industry. Specification Group (ISG) Zero-Touch Network and Service Management (ZSM), while (ii) keeping the ACROSS
architecture vendor and technology-agnostic.
Deep end-to-end system telemetry
to unlock visibility of the intertwined network and compute system states. ACROSS overcomes the key bottlenecks of today’s out-of-band monitoring systems fostering an in-band telemetry approach which achieves nano-to-microsecond-level system (i.e., joint network and compute) state visibility at the speed of the hardware, via programmable telemetry offloaded into modern open programmable hardware.
AI-driven intelligence
that leverages both the vast number of events originating from the dense IoT ecosystem and detailed monitoring data originating from the ACROSS telemetry sub-system, to generate new insights – even without previously acquired knowledge – on the current and future network and system states, in the form of quick and accurate detections, diagnoses, and predictions that enrich the control loops of the orchestration platform, both at the level of a single domain as well as across multiple domains.
Full-stack cross-domain zero-touch provisioning
via programmable “hooks” that exploit events originating both from the raw telemetry and device planes (Axis #3) as well as events produced by the ACROSS AI stratum (Axis #4), to invoke automated fully-unsupervised orchestration operations (Axis #1) in response to these events.
Secure and trusted orchestration
is more important than ever as ACROSS needs to reason not only about the trustworthiness of the massive IoTs and user devices, but also protect the vast data volumes generated by these devices, while ensuring that the overlay AI engine runs in a trusted environment, thus producing trustworthy intelligence.
Test-case driven methodology
that follows an iterative approach starting from (i) a solid taxonomy of zero-touch operations originating from the leading role & experience of key ACROSS partners in standardisation efforts (e.g., TID leading ETSI ZSM), then (ii) mapping of this activity to technical features undergoing an agile lifecycle development, and (iii) targeted demonstrations over real geo-distributed testbeds, targeting 6G KPIs and recently-defined KVIs, and (iv) mapping to real-life use cases benefiting the society, environment, and industry.
Drive standardisation and catalyse innovation
in zero-touch network and service management through tangible prototyping, experimentation, and demonstration activities, including (i) proof of concept (PoC) project submissions to ETSI ISG ZSM and (ii) demonstrable open-source contributions to ETSI NFV Open-Source MANO (OSM) at OSM hackfests or through the OSM PoC framework. These efforts will be tightly aligned with the latest standardisation and development activities of the community, aiming at (i) validating the ACROSS compliance with relevant initiatives but more importantly (ii) render ACROSS an exemplar platform that will champion future standardisation efforts, thus shape the future of service deployment and orchestration systems.