Flexible Wideband Filtering Monopole Antenna With Stable High Omnidirectionality

This article presents a wideband flexible filtering monopole antenna with symmetric structure for stable high omnidirectionality. It is based on a monopole antenna, which is printed on a single-layer flexible substrate. Two folded parasitic strips with different length are devised on both sides of the driven monopole, giving filtering responses in the higher and lower band without filtering circuits. Since the asymmetric filtering structure adversely affects in-band omnidirectionality, this baseline design is extended with symmetric filtering structure to improve omnidirectionality and bandwidth. In the proposed design, a pair of parasitic strips are devised on the both sides of monopole antenna symmetrically, achieving a radiation null in the higher band. Then, by loading a pair of folded parasitic strips on the both sides of feed line with slotted metal ground, a radiation null is realized in the lower band. Besides, the driven monopole is slotted symmetrically for wideband operation. By adopting a fully symmetric filtering structure, the proposed design effectively suppresses the impact of the parasitic elements on the in-band omnidirectional radiation pattern, thereby achieving high omnidirectionality. Furthermore, the proposed antenna exhibits stable performance under different bending radii. To verify our design concept, an antenna prototype is fabricated. Both the flat and bent antennas are measured. The results show that the proposed antenna has a -10 dB impedance bandwidth of 45.6%, an in-band gain about 2 dBi, and an out-of-band radiation suppression more than 11 dB. The measured omnidirectionality has variations less than 0.8 dB without bending and 1 dB with a bending radius of 30 mm. This design offers several advantages including stable high omnidirectionality across a wide bandwidth, flexible conformal capability, and filtering property.

Flexible Wearable Filtering Antenna With Stable Performance for IoT Devices

In this article, a flexible and lightweight filtering wearable antenna without extra circuits is presented. The proposed antenna starts from a flexible directional antenna with lightweight structure, which includes a layer of ultra-thin flexible substrate, a metal ground layer, and a flexible foam layer sandwiched between them. Then, by introducing two pairs of vertical slots to the radiation patch printed on the flexible substrate, two radiation nulls are realized at both band edges without extra circuits. Moreover, to mitigate the deterioration of in-band radiation under different curvature, a pair of inverted Fshaped slots are loaded on the radiation patch. The coupling of Fshaped slots suppresses non-radiated lateral current components along the curvature direction, maintaining stable performance after bending. In addition, deformation analysis of the proposed antenna with a three-layer human tissue model under different bending radii is carefully carried out, showing stable bandwidth, effective out-of-band radiation suppression, and low specific absorption rate (SAR) value. To verify this method, a prototype is fabricated. Measurements are conducted both in free space and conformal on the curved body tissue. The results show that the proposed antenna achieves a bandwidth from 2.7 GHz to 3 GHz, an out-of-band radiation suppression more than 11 dB with maxmium suppression of 23 dB, and an average gain of 8.5 dBi. As a flexible wearable antenna with stable performance and integrated reliable filtering features, it has several advantages including flexible wearable structure, stable filtering properties, lightweight characteristic, and low SAR. This makes it an excellent candidate for wearable IoT applications.

M*: A Modular, Extensible, Serving System for Multimodal Models

We are entering a new era of composite model architectures that integrate diverse components such as vision encoders, language backbones, diffusion and flow heads, audio codecs, action generators, and world-model predictors. Such architectures underpin a broad class of multimodal models, including unified multimodal models, omni models, speech-language models, vision-language-action policies, and world models. However, existing model serving frameworks were built on narrow assumptions about model structure, making them ill-suited to accommodate this new architectural diversity. Here we present M*, a universal serving system for efficient serving of composite AI models. M* represents models as dataflow graphs, processing requests spanning diverse modalities and tasks as traversals over these graphs. The core insight is a modular abstraction that supports arbitrary composition of model components, flexible placement onto a physical cluster, and model-agnostic optimizations within a distributed runtime. We call this abstraction the Walk Graph and show how it can concisely capture composite models from a broad range of families. We instantiate M* on representative models and find that it achieves, on average, 20% lower end-to-end latency than vLLM-Omni for text-to-image workloads on BAGEL, while delivering up to 2.9x lower real-time factor and 2.7x higher throughput for text-to-speech workloads on Qwen3-Omni. M* also outperforms the V-JEPA 2-AC rollout baseline for robotic planning by up to 12.5x. Thus, our work paves the road towards more efficient serving of complex models with minimal developer effort.