英国SPF招标：频谱管理人工智能

Invitation to Tender (ITT) Independent study ontheAIfor spectrum management 01August2024 Background In 2023,DSIT issued a statement outlining the relevance of AI and ML techniques tomitigate interference and optimize the useof spectrum. With governmental supporttoAIadoption, commercial interestaroundthis trendsupportsthe provision of smart solutions forspectrum management. Theincreasing demand for spectrumposeschallenges to traditional spectrum managementmethods. Innovative approachesalready seek to increaseefficient spectrumallocation andutilization in certain conditions, such as dynamic spectrum access (DSA).Within thatcontext, AIhas the potential tofurtherimprovefrequencycoordination,propagationestimation, andultimatelydeliver moreefficient spectrum utilisation. AIadoption bynational regulatory agenciesstill merits deeper studies.Thepotentialefficiencygains forspectrummanagementshould be measured againstconsiderationsaround costs and risks.Without adding further general definitions on artificial intelligenceand re-stating on the overarching benefits of implementing machine learningto existingprocesses, the current project aims to outline specific developments to spectrummanagement from the introduction of AI/ML. Scope Theresearchproject will seek to understand how AI could add efficiency to spectrummanagement in the UK landscape and allow for improvedsharingof frequency bands.TheUK SPF aimsto understand the impacts of the introduction of thesesolutionsto existinglicensing processes, anticipating thewideradoption of AI by 6G networks.The latter couldhave regulatory and policy implications on future authorisation of 6G networks. The study needs to comprehend the perspective of therelevant parties. It should alsoconsider the different technologies and related processes that will be required forthedeployment of AI for spectrum management. Particularlyrelevant is to identify the elementsto conduct acost-benefitanalysisaround the trade-offs for implementation byOfcom. Thestudy should considerthepotentialcosts tostakeholdersandthemodel forrisk allocationassociated with the deployment of thistechnologyas well as the responsibilities arising fromimplementing such technology in the licencing and spectrum management processes. Another aspect is the collaboration required to achieve implementation.On the commercialside,there are severalquestionsregardingthe business case to launchthese solutions.There are data and intellectual property considerations that need to be weighed against thepotential adoptionof processesfor spectrum managementthat are based on AI/MLapproaches. It also needs to consider the costs it would add torelevant partiesto guaranteeresilience and security. The appointed consultant isexpected to map trials and objectives outlined by othercountries. At the same time,the study shouldexplorethe capabilities that are unique to theUK.This perspective will shed light on the ability ofthe UKregulatortounderstand thesolutions available, as well as the riskand responsibilitystructure associated with its implementation.It will also outlineeventual returns in capital through efficient andinterference-free use of spectrumto society, whichwillhelpdetermine the type of the AIalgorithms applicable to spectrum management. Expected deliverables 1.A report detailing thefindings including:a.Mapinternationaltrials and objectivesdevisedby other countriesfor the use of AI technologies in spectrum management.b.Identifythepotential AI technologiesthatfitto the UKcontextalong with aroadmap to theintroduction of these technologies to existing licensing processes,anticipatingfor examplethe large adoption of AI by 6G networks.Elaborate on therequirements of those technologies in terms of what operational or other data needsto be captured, collected and kept by the algorithms to provide improvements to thespectrum management process.c.Analysis oftheconsequences from thedifferent technology solutions,commercialviability,relatedpolicies,associatedrisksandresponsibilitiesas well as the extent ofpotential collaboration needed amongthe involved parties (from an End-to-Endperspective).d.Pros and consofthe adoption ofAIalgorithmstospectrummanagementbytheregulatorin relation to theefficiency gains, costs to stakeholders/regulatorsandrelated risks. 2.Additionalevidence toperform acost-benefit analysisof adopting AI tospectrummanagementoutlining thepotential collaboration between relevant partieswithinthe currentregulatoryframework. Cost: •Total budget is of the order of £25,000 (+VAT) Timetable: •ITT issued:01August2024•Deadline for clarification of questions1:22August2024•Deadline to submit your tender:04September2024 Duration: •Expected duration of the study: approximatively three months, including deliveringthe report.•To enable transparency and more efficient delivery management, it is suggested, atthe time ofcommencement of the report, to establish monthly checkpoint meetings,during which UK SPF could provide u