Electronic voting has partially solved the problems of poor anonymity and low efficiency associated with traditional voting.However,the difficulties it introduces into the supervision of the vote counting,as well as i...Electronic voting has partially solved the problems of poor anonymity and low efficiency associated with traditional voting.However,the difficulties it introduces into the supervision of the vote counting,as well as its need for a concurrent guaranteed trusted third party,should not be overlooked.With the advent of blockchain technology in recent years,its features such as decentralization,anonymity,and non-tampering have made it a good candidate in solving the problems that electronic voting faces.In this study,we propose a multi-candidate voting model based on the blockchain technology.With the introduction of an asymmetric encryption and an anonymity-preserving voting algorithm,votes can be counted without relying on a third party,and the voting results can be displayed in real time in a manner that satisfies various levels of voting security and privacy requirements.Experimental results show that the proposed model solves the aforementioned problems of electronic voting without significant negative impact from an increasing number of voters or candidates.展开更多
Speculative decoding accelerates Large Language Model(LLM)inference by allowing a lightweight draft model to predict multiple future tokens that are subsequently verified by a larger target model.In AI-native Radio Ac...Speculative decoding accelerates Large Language Model(LLM)inference by allowing a lightweight draft model to predict multiple future tokens that are subsequently verified by a larger target model.In AI-native Radio Access Networks(AI-RAN),this mechanism naturally enables device-edge collaborative inference.However,existing distributed speculative decoding schemes incur significant uplink communication overhead,as they require transmitting full-vocabulary logits at every decoding step.To address this challenge,we propose a sparsify-then-sample strategy,termed Truncated Sparse Logits Transmission(TSLT),which transmits only the logits and indices of a truncated candidate set.We provide theoretical guarantees showing that TSLT preserves the acceptance rate of speculative decoding.The proposed framework is further extended to a multi-candidate setting,where multiple draft candidates per step increase the acceptance probability.Extensive experiments demonstrate that TSLT substantially reduces uplink communication while maintaining end-to-end inference latency and model quality,validating its effectiveness for scalable and communication-efficient distributed LLM inference in future AI-RAN systems.展开更多
基金This work was supported in part by Shandong Provincial Natural Science Foundation(ZR2019PF007)the National Key Research and Development Plan of China(2018YFB0803504)+2 种基金Basic Scientific Research Operating Expenses of Shandong University(2018ZQXM004)Guangdong Province Key Research and Development Plan(2019B010137004)the National Natural Science Foundation of China(U20B2046).
文摘Electronic voting has partially solved the problems of poor anonymity and low efficiency associated with traditional voting.However,the difficulties it introduces into the supervision of the vote counting,as well as its need for a concurrent guaranteed trusted third party,should not be overlooked.With the advent of blockchain technology in recent years,its features such as decentralization,anonymity,and non-tampering have made it a good candidate in solving the problems that electronic voting faces.In this study,we propose a multi-candidate voting model based on the blockchain technology.With the introduction of an asymmetric encryption and an anonymity-preserving voting algorithm,votes can be counted without relying on a third party,and the voting results can be displayed in real time in a manner that satisfies various levels of voting security and privacy requirements.Experimental results show that the proposed model solves the aforementioned problems of electronic voting without significant negative impact from an increasing number of voters or candidates.
基金supported by National Key Research and Development Program of China(2024YFE0200800)in part by Major Key Project of PCL(PCL2025AS209)in part by Guangdong S&T Programme(2024B0101010003).
文摘Speculative decoding accelerates Large Language Model(LLM)inference by allowing a lightweight draft model to predict multiple future tokens that are subsequently verified by a larger target model.In AI-native Radio Access Networks(AI-RAN),this mechanism naturally enables device-edge collaborative inference.However,existing distributed speculative decoding schemes incur significant uplink communication overhead,as they require transmitting full-vocabulary logits at every decoding step.To address this challenge,we propose a sparsify-then-sample strategy,termed Truncated Sparse Logits Transmission(TSLT),which transmits only the logits and indices of a truncated candidate set.We provide theoretical guarantees showing that TSLT preserves the acceptance rate of speculative decoding.The proposed framework is further extended to a multi-candidate setting,where multiple draft candidates per step increase the acceptance probability.Extensive experiments demonstrate that TSLT substantially reduces uplink communication while maintaining end-to-end inference latency and model quality,validating its effectiveness for scalable and communication-efficient distributed LLM inference in future AI-RAN systems.
