ABSTRACT In federated learning (FL)‐assisted Internet of Things (IoT) systems, FL trains models using datasets on various client devices without sending the datasets to a centralized server. This approach enhances the accuracy and reliability of models while preserving the privacy of client devices. However, FL implementations face challenges, such as single point of server failure and lack of incentives. To address the server failure issue, a backup server can be added. Meanwhile, each FL client has varying data quality and motivations to participate, leading to differences in the quality of local models uploaded to the server. To motivate clients to contribute more, we designed a novel incentive mechanism based on the Stackelberg game. This mechanism allocates rewards based on the quality of the models each client uploads, rather than the amount of data trained. We separately modelled the utilities of the server and the clients, allowing the server to rationally allocate rewards based on each client's contribution to model training. After analysing the utilities, we transform the game into two optimization problems and develop an algorithm whose per‐round complexity scales linearly with the number of clients under fixed numerical tolerances. The obtained equilibrium matches exhaustive search within numerical precision while significantly reducing computation.

Positioning and sensing, as critical enablers for emerging applications, can be further empowered by multi-band fusion technology. In this paper, a two-stage framework utilizing position-domain joint stochastic particle variational Bayesian inference (PD-JSPVBI) is proposed to address the key challenges in time-of-arrival (TOA)-based direct position determination. Unlike existing works focusing on multi-band delay estimation or single-band direct positioning, our unified framework enables joint target localization directly from multi-station, multiband signals. The algorithm integrates prior information and resolves spatial coordinate coupling via a novel two-dimensional particle-based variational approximation, significantly improving estimation efficiency. Additionally, a joint estimation mechanism is introduced to synchronously optimize positioning parameters (e.g., target coordinates) and non-ideal factors (e.g., timing synchronization errors, random initial phases) within a unified variational framework. Simulation results validate the proposed algorithm's superiority, outperforming conventional cascaded architectures by eliminating error propagation and leveraging multi-band coherence. The proposed solution offers a promising pathway for high-precision direct positioning systems.

MOTIVATION: Tumor polyvalent neoantigen mRNA vaccines are gaining prominence in immunotherapy. The design of sequences in vaccine development is crucial for enhancing both the immunogenicity and safety of vaccines. However, a major challenge lies in selecting the optimal sequences from the large pools generated by multiple peptide combinations and synonymous codons. RESULTS: We introduce NeoDesign, a computational tool designed to tackle the challenge of sequence design. NeoDesign comprises four modules: Library Construction, Optimal Path Filtering, Linker Addition, and λ-Evaluation. It aims to identify the optimal protein sequence for tumor polyvalent neoantigen vaccines by minimizing linker usage, avoiding unexpected neoantigens and functional domains, and simplifying the structure. It also provides a preference scheme to balance mRNA stability and protein expression when designing mRNA sequences for the optimal protein sequence. This tool can potentially improve the sequence design of tumor polyvalent neoantigen mRNA vaccines, thereby significantly advancing immunotherapy strategies. AVAILABILITY AND IMPLEMENTATION: NeoDesign is freely available on https://github.com/HuangLab-Fudan/neoDesign and https://figshare.com/projects/NeoDesign/221704.

Multiband fusion enhances delay estimation by jointly utilizing signals from multiple noncontiguous frequency bands. However, in the multiband signal model, there are many local optimums in the associated likelihood function due to the existence of high-frequency component and phase distortion factors, posing challenges for high-accuracy parameter estimation. To address this, we propose a two-stage scheme equipped with different signal models derived from the original model, where the first-stage coarse estimation is performed using a weighted root MUSIC algorithm to narrow down the search range for the subsequent stage, and the second-stage refined estimation utilizes a Bayesian approach to avoid convergence to bad suboptimal solutions. Specifically, we apply the block stochastic successive convex approximation (SSCA) approach to derive a novel stochastic particle-based variational Bayesian inference (SPVBI) algorithm in the refined stage. Unlike conventional particle-based VBI (PVBI) that optimizes only particle probability and incurs exponential per-iteration complexity with particle count, our more flexible SPVBI algorithm optimizes both the position and probability of each particle. Additionally, it utilizes block SSCA to significantly improve sampling efficiency by averaging over iterations, making it suitable for high-dimensional problems. Extensive simulations demonstrate the superiority of our proposed algorithm over various baseline methods.

Future wireless networks are envisioned to provide ubiquitous sensing services, driving a substantial demand for multi-dimensional non-convex parameter estimation. This entails dealing with non-convex likelihood functions containing numerous local optima. Variational Bayesian inference (VBI) provides a powerful tool for modeling complex estimation problems and leveraging prior information, but poses a long-standing challenge on computing intractable posterior distributions. Most existing variational methods depend on specific distribution assumptions for obtaining closed-form solutions, and are difficult to apply in practical scenarios. Given these challenges, firstly, we propose a parallel stochastic particle VBI (PSPVBI) algorithm. Due to innovations like particle approximation, added updates of particle positions, and parallel stochastic successive convex approximation (PSSCA), PSPVBI can flexibly drive particles to fit the posterior distribution with acceptable complexity, yielding high-precision estimates of the target parameters. Furthermore, additional speedup can be obtained by deep-unfolding this algorithm. Specifically, superior hyperparameters are learned to dramatically reduce iterations. In this PSPVBI-induced deep-unfolding network, some techniques related to gradient computation, data sub-sampling, differentiable sampling, and generalization ability are also employed to facilitate the practical deployment. Finally, we apply the learnable PSPVBI (LPSPVBI) to solve two important positioning/sensing problems over wireless networks. Simulations indicate that the LPSPVBI algorithm outperforms existing solutions.

In integrated sensing and communication (ISAC) systems, communication signals are exploited to achieve high-accuracy sensing. Multiband Wi-Fi sensing, which jointly utilizes Wi-Fi signals from multiple non-contiguous frequency bands to improve the sensing performance, has recently emerged as a promising technology for ISAC. However, the multi-dimensional non-convex likelihood function associated with the multiband WiFi sensing contains many local optimums due to the existence of high frequency components and phase distortion factors in the signal model, making it difficult to exploit the multiband gain for high-accuracy parameter estimation. To address this, we divide the target parameter estimation into two stages equipped with different signal models derived from the original model, where the first-stage coarse estimation is used to narrow down the search range for the next stage, and the second-stage refined estimation is based on the Bayesian approach to avoid the convergence to a bad local optimum of the likelihood function. Specifically, we apply the block stochastic successive convex approximation (SSCA) approach to derive a novel stochastic particle-based variational Bayesian inference (SPVBI) algorithm in the refined stage. Unlike the conventional particle-based VBI (PVBI) in which only particle probability is optimized and the per-iteration computational complexity increases exponentially with particle count, the proposed SPVBI optimizes both the position and probability of each particle, and it adopts the block SSCA to significantly improve the sampling efficiency by averaging over iterations. As such, the proposed SPVBI can achieve a better performance than the conventional PVBI with a much lower complexity. Finally, simulations verify the advantage of the proposed algorithm over various baseline algorithms.