Generative Recommenders (GRs) have recently emerged as promising alternatives to traditional Deep Learning Recommendation Models (DLRMs). Despite their potential, GRs remain computationally expensive in inference, exhibiting compute-bound characteristics similar to the prefill stage of Large Language Model (LLM) inference. Prefix caching can reduce redundant computation by reusing previously constructed KV caches. However, the unique properties of GRs, i.e., highly personalized user profiles and real-time item retrieval, make cache reuse across queries challenging, resulting in limited computational savings.

In this study, the authors study the max–min fairness for wireless energy transfer in a multiuser multiple‐input multiple‐output communication system with simultaneous wireless information and power transfer. In particular, they aim to maximise the minimum harvested energy among the multiple multi‐antenna energy receivers while guaranteeing secure communication for multi‐antenna information receiver. The dual use of artificial noise to facilitate both wireless energy transfer and secure communication is exploited in the authors’ proposed problem. Both scenarios of perfect and imperfect channel state information (CSI) known at the transmitter are considered. For the perfect CSI case, the formulated max–min energy harvesting (MM‐EH) problem is non‐convex and intractable. To circumvent it, an iterative optimisation algorithm based on Taylor series expansion is proposed. Then, they turn their attention to the imperfect CSI case, where a max–min robust energy harvesting (MMR‐EH) problem is considered. Though the MMR‐EH problem is more complicated than the MM‐EH problem, they reveal that the iterative optimisation method can be extended to the solution of the former, wherein the S‐procedure is introduced. Simulation results show the efficiency of their proposed solutions in terms of energy harvesting.

In this letter, we consider a multiple-input-multiple-output (MIMO) downlink system with simultaneous wireless information and power transfer (SWIPT), where the information sent to the desired receiver (DR) may be wiretapped by the malicious energy harvesting (EH) receivers (potential eavesdroppers). Assuming the channel state information (CSI) of the EH receivers is not perfectly known at the transmitter, we aim to maximize the worst-case secrecy rate by jointly designing the precoding matrix, AN covariance matrix and power splsplitting ratio, under the constraints of total transmit power and harvested energyitting ratio, under the constraints of total transmit power and harvested energy at receivers. The formulated problem is non-convex and semi-infinite, which is hard to tackle. To solve it, we first deal with the non-convexity caused by CSI errors with the aid of the S-Procedure, and then employ the Taylor series approximation techniques to transfer the nonconvex problem into a convex optimization problem. Based on that, we propose an iterative algorithm with proved convergence to solve the original problem. Simulation results are finconvex optimization problem. Based on that, we propose an iterative algorithmally provided to show the effectiveness of our proposed robust transmit design.