We address the challenge of representing long captions in vision-language models, such as CLIP. By design these models are limited by fixed, absolute positional encodings, restricting inputs to a maximum of 77 tokens and hindering performance on tasks requiring longer descriptions. Although recent work has attempted to overcome this limit, their proposed approaches struggle to model token relationships over longer distances and simply extend to a fixed new token length. Instead, we propose a generalizable method, named TULIP, able to upgrade the token length to any length for CLIP-like models. We do so by improving the architecture with relative position encodings, followed by a training procedure that (i) distills the original CLIP text encoder into an encoder with relative position encodings and (ii) enhances the model for aligning longer captions with images. By effectively encoding captions longer than the default 77 tokens, our model outperforms baselines on cross-modal tasks such as retrieval and text-to-image generation. The code repository is available at https://github.com/ivonajdenkoska/tulip.
Low-Dimension-to-High-Dimension (LDHD) generalization is a special case of Out-of-Distribution (OOD) generalization, where the training data are restricted to a low-dimensional subspace of the high-dimensional testing space. Assuming that each instance is generated from a latent variable and the dimension of the latent variable reflects the problem scale, the inherent scaling challenge in length generalization can be captured by the LDHD generalization in the latent space. We theoretically demonstrate that LDHD generalization is generally unattainable without exploiting prior knowledge to provide appropriate inductive bias. Specifically, we explore LDHD generalization in Boolean functions. We verify that different architectures trained with (S)GD converge to \emph{min-degree interpolators w.r.t. different independent sets}. LDHD generalization is achievable if and only if the target function coincides with this inductive bias. Applying the insights from LDHD generalization to length generalization, we explain the effectiveness of CoT as changing the structure latent space to enable better LDHD generalization. We also propose a principle for position embedding design to handle both the inherent LDHD generalization and the nuisances such as the data format. Following the principle, we propose a novel position embedding called RPE-Square that remedies the RPE for dealing with the data format nuisance.
Length extrapolation algorithms based on Rotary position embedding (RoPE) have shown promising results in extending the context length of language models. However, understanding how position embedding can capture longer-range contextual information remains elusive. Based on the intuition that different dimensions correspond to different frequency of changes in RoPE encoding, we conducted a dimension-level analysis to investigate the correlation between a hidden dimension of an attention head and its contribution to capturing long-distance dependencies. Using our correlation metric, we identified a particular type of attention heads, which we named Positional Heads, from various length-extrapolated models. These heads exhibit a strong focus on long-range information interaction and play a pivotal role in long input processing, as evidence by our ablation. We further demonstrate the correlation between the efficiency of length extrapolation and the extension of the high-dimensional attention allocation of these heads. The identification of Positional Heads provides insights for future research in long-text comprehension.
This paper proposes a fluid antenna (FA)-assisted near-field integrated sensing and communications (ISAC) system enabled by the extremely large-scale simultaneously transmitting and reflecting surface (XL-STARS). By optimizing the communication beamformer, the sensing signal covariance matrix, the XL-STARS phase shift, and the FA position vector, the Cram\'er-Rao bound (CRB), as a metric for sensing performance, is minimized while ensuring the standard communication performance. A double-loop iterative algorithm based on the penalty dual decomposition (PDD) and block coordinate descent (BCD) methods is proposed to solve the non-convex minimization problem by decomposing it into three subproblems and optimizing the coupling variables for each subproblem iteratively. Simulation results validate the superior performance of the proposed algorithm.
Movable antenna (MA) is an emerging technology that can significantly improve communication performance via the continuous adjustment of the antenna positions. To unleash the potential of MAs in wideband communication systems, acquiring accurate channel state information (CSI), i.e., the channel frequency responses (CFRs) between any position pair within the transmit (Tx) region and the receive (Rx) region across all subcarriers, is a crucial issue. In this paper, we study the channel estimation problem for wideband MA systems. To start with, we express the CFRs as a combination of the field-response vectors (FRVs), delay-response vector (DRV), and path-response tensor (PRT), which exhibit sparse characteristics and can be recovered by using a limited number of channel measurements at selected position pairs of Tx and Rx MAs over a few subcarriers. Specifically, we first formulate the recovery of the FRVs and DRV as a problem with multiple measurement vectors in compressed sensing (MMV-CS), which can be solved via a simultaneous orthogonal matching pursuit (SOMP) algorithm. Next, we estimate the PRT using the least-square (LS) method. Moreover, we also devise an alternating refinement approach to further improve the accuracy of the estimated FRVs, DRV, and PRT. This is achieved by minimizing the discrepancy between the received pilots and those constructed by the estimated CSI, which can be efficiently carried out by using the gradient descent algorithm. Finally, simulation results demonstrate that both the SOMP-based channel estimation method and alternating refinement method can reconstruct the complete wideband CSI with high accuracy, where the alternating refinement method performs better despite a higher complexity.
Reconfigurable intelligent surfaces (RISs) are key enablers for integrated sensing and communication (ISAC) systems in the 6G communication era. With the capability of dynamically shaping the channel, RISs can enhance communication coverage. Additionally, RISs can serve as additional anchors with high angular resolution to improve localization and sensing services in extreme scenarios. However, knowledge of anchors' states such as position, orientation, and hardware impairments are crucial for localization and sensing applications, requiring dedicated calibration, including geometry and hardware calibration. This paper provides an overview of various types of RIS calibration, their impacts, and the challenges they pose in ISAC systems.
In mobile systems with low-altitude vehicles, integrated sensing and communication (ISAC) is considered an effective approach to increase the transmission rate due to limited spectrum resources. To further improve the ISAC performance, this paper proposes a novel method called integrated sensing and communication-movable antenna (ISAC-MA) to optimize the antenna's position. Our goal is to support low-space vehicles by optimizing radar and communication joint beamforming and antenna position in the presence of clutter. This scheme not only guarantees the required signal-to-noise ratio (SNR) for sensing but also further improves the SNR for communication. A successive convex approximation (SCA)-based block coordinate descent (BCD) algorithm is proposed to maximize communication capacity under the condition of sensing SNR. Numerical results show that, compared with the traditional ISAC system and various benchmark schemes, the proposed ISAC-MA system can achieve higher communication capacity under the same sensing SNR constraints.
For millimeter-wave (mmWave) non-orthogonal multiple access (NOMA) communication systems, we propose an innovative near-field (NF) transmission framework based on dynamic metasurface antenna (DMA) technology. In this framework, a base station (BS) utilizes the DMA hybrid beamforming technology combined with the NOMA principle to maximize communication efficiency between near-field users (NUs) and far-field users (FUs). In conventional communication systems, obtaining channel state information (CSI) requires substantial pilot signals, significantly reducing system communication efficiency. We propose a beamforming design scheme based on position information to address with this challenge. This scheme does not depend on pilot signals but indirectly obtains CSI by analyzing the geometric relationship between user position information and channel models. However, in practical applications, the accuracy of position information is challenging to guarantee and may contain errors. We propose a robust beamforming design strategy based on the worst-case scenario to tackle this issue. Facing with the multi-variable coupled non-convex problems, we employ a dual-loop iterative joint optimization algorithm to update beamforming using block coordinate descent (BCD) and derive the optimal power allocation (PA) expression. We analyze its convergence and complexity to verify the proposed algorithm's performance and robustness thoroughly. We validate the theoretical derivation of the CSI error bound through simulation experiments. Numerical results show that our proposed scheme performs better than traditional beamforming schemes. Additionally, the transmission framework exhibits strong robustness to NU and FU position errors, laying a solid foundation for the practical application of mmWave NOMA communication systems.
In future 6G networks, anti-jamming will become a critical challenge, particularly with the development of intelligent jammers that can initiate malicious interference, posing a significant security threat to communication transmission. Additionally, 6G networks have introduced mobile edge computing (MEC) technology to reduce system delay for edge user equipment (UEs). Thus, one of the key challenges in wireless communications is minimizing the system delay while mitigating interference and improving the communication rate. However, the current fixed-position antenna (FPA) techniques have limited degrees of freedom (DoF) and high power consumption, making them inadequate for communication in highly interfering environments. To address these challenges, this paper proposes a novel MEC anti-jamming communication architecture supported by mobile antenna (MA) technology. The core of the MA technique lies in optimizing the position of the antennas to increase DoF. The increase in DoF enhances the system's anti-jamming capabilities and reduces system delay. In this study, our goal is to reduce system delay while ensuring communication security and computational requirements. We design the position of MAs for UEs and the base station (BS), optimize the transmit beamforming at the UEs and the receive beamforming at the BS, and adjust the offloading rates and resource allocation for computation tasks at the MEC server. Since the optimization problem is a non-convex multi-variable coupled problem, we propose an algorithm based on penalty dual decomposition (PDD) combined with successive convex approximation (SCA). The simulation results demonstrate that the proposed MA architecture and the corresponding schemes offer superior anti-jamming capabilities and reduce the system delay compared to FPA.
Cellular-connected unmanned aerial vehicle (UAV) communications is an enabling technology to transmit control signaling or payload data for UAVs through cellular networks. Due to the line-of-sight (LoS) dominant air-to-ground channels, efficient interference mitigation is crucial to UAV communications, while the conventional fixed-position antenna (FPA) arrays have limited degrees of freedom (DoFs) to suppress the interference between the UAV and its non-associated co-channel base stations (BSs). To address this challenge, we propose in this letter a new approach by utilizing the six-dimensional movable antenna (6DMA) arrays to enhance the interference mitigation for the UAV. Specifically, we propose an efficient block coordinate descent (BCD) algorithm to iteratively optimize the antenna position vector (APV), array rotation vector (ARV), receive beamforming vector, and associated BS of the UAV to maximize its signal-to-interference-plus-noise ratio (SINR). Numerical results show that the proposed 6DMA enhanced cellular-connected UAV communication can significantly outperform that with the traditional FPA arrays and other benchmark schemes in terms of interference mitigation.
As a physical layer security technology, directional modulation (DM) can be combined with intelligent reflect-ing surface (IRS) to improve the security of drone communications. In this paper, a directional modulation scheme assisted by the IRS is proposed to maximize the transmission rate of unmanned aerial vehicle (UAV) secure communication. Specifically, with the assistance of the IRS, the UAV transmits legitimate information and main-tains its constellation pattern at the location of legitimate users on the ground, while the constellation pattern is disrupted at the eavesdropper's location. In order to solve the joint optimization problem of digital weight coefficients, UAV position, and IRS discrete phase shift, firstly, the digital weight vector and UAV position are optimized through power minimization. Secondly, three methods are proposed to optimize IRS phase shift, namely vector trajectory (VT) method, cross entropy vector trajectory (CE-VT) algorithm, and block coordinate descent vector trajectory (BCD-VT) algorithm. Compared to traditional cross entropy (CE) methods and block coordinate descent (BCD) methods, the proposed CE-VT and BCD-VT algorithms can improve transmission rate performance. The numerical results validate the effectiveness of the optimization scheme in IRS assisted UAV communication.
Gain-phase impairments (GPIs) affect both communication and sensing in 6G integrated sensing and communication (ISAC). We study the effect of GPIs in a single-input, multiple-output orthogonal frequency-division multiplexing ISAC system and develop a model-based unsupervised learning approach to simultaneously (i) estimate the gain-phase errors and (ii) localize sensing targets. The proposed method is based on the optimal maximum a-posteriori ratio test for a single target. Results show that the proposed approach can effectively estimate the gain-phase errors and yield similar position estimation performance as the case when the impairments are fully known.
Intelligent reflecting surface (IRS) and movable antenna (MA) technologies have been proposed to enhance wireless communications by creating favorable channel conditions. This paper investigates the joint beamforming and antenna position design for an MA-enabled IRS (MA-IRS)-aided multi-user multiple-input single-output (MU-MISO) communication system, where the MA-IRS is deployed to aid the communication between the MA-enabled base station (BS) and user equipment (UE). In contrast to conventional fixed position antenna (FPA)-enabled IRS (FPA-IRS), the MA-IRS enhances the wireless channel by controlling the positions of the reflecting elements. To verify the system's effectiveness and optimize its performance, we formulate a sum-rate maximization problem with a minimum rate threshold constraint for the MU-MISO communication. To tackle the non-convex problem, a product Riemannian manifold optimization (PRMO) method is proposed for the joint design of the beamforming and MA positions. Specifically, a product Riemannian manifold space (PRMS) is constructed and the corresponding Riemannian gradient is derived for updating the variables, and the Riemannian exact penalty (REP) method and a Riemannian Broyden-Fletcher-Goldfarb-Shanno (RBFGS) algorithm is derived to obtain a feasible solution over the PRMS. Simulation results demonstrate that compared with the conventional FPA-IRS-aided MU-MISO communication, the reflecting elements of the MA-IRS can move to the positions with higher channel gain, thus enhancing the system performance. Furthermore, it is shown that integrating MA with IRS leads to higher performance gains compared to integrating MA with BS.
Movable antennas (MAs), which can be swiftly repositioned within a defined region, offer a promising solution to the limitations of fixed-position antennas (FPAs) in adapting to spatial variations in wireless channels, thereby improving channel conditions and communication between transceivers. However, frequent MA position adjustments based on instantaneous channel state information (CSI) incur high operational complexity, making real-time CSI acquisition impractical, especially in fast-fading channels. To address these challenges, we propose a two-timescale transmission framework for MA-enabled multiuser multiple-input-multiple-output (MU-MIMO) systems. In the large timescale, statistical CSI is exploited to optimize MA positions for long-term ergodic performance, whereas, in the small timescale, beamforming vectors are designed using instantaneous CSI to handle short-term channel fluctuations. Within this new framework, we analyze the ergodic sum rate and develop efficient MA position optimization algorithms for both maximum-ratio-transmission (MRT) and zero-forcing (ZF) beamforming schemes. These algorithms employ alternating optimization (AO), successive convex approximation (SCA), and majorization-minimization (MM) techniques, iteratively optimizing antenna positions and refining surrogate functions that approximate the ergodic sum rate. Numerical results show significant ergodic sum rate gains with the proposed two-timescale MA design over conventional FPA systems, particularly under moderate to strong line-of-sight (LoS) conditions. Notably, MA with ZF beamforming consistently outperforms MA with MRT, highlighting the synergy between beamforming and MAs for superior interference management in environments with moderate Rician factors and high user density, while MA with MRT can offer a simplified alternative to complex beamforming designs in strong LoS conditions.
We present a new guaranteed-safe model predictive path integral (GS-MPPI) control algorithm that enhances sample efficiency in nonlinear systems with multiple safety constraints. The approach use a composite control barrier function (CBF) along with MPPI to ensure all sampled trajectories are provably safe. We first construct a single CBF constraint from multiple safety constraints with potentially differing relative degrees, using it to create a safe closed-form control law. This safe control is then integrated into the system dynamics, allowing MPPI to optimize over exclusively safe trajectories. The method not only improves computational efficiency but also addresses the myopic behavior often associated with CBFs by incorporating long-term performance considerations. We demonstrate the algorithm's effectiveness through simulations of a nonholonomic ground robot subject to position and speed constraints, showcasing safety and performance.
Doppler ultrasound is widely used to monitor fetal heart rate during labor and pregnancy. Unfortunately, it is highly sensitive to fetal and maternal movements, which can cause the displacement of the fetal heart with respect to the ultrasound beam, in turn reducing the Doppler signal-to-noise ratio and leading to erratic, noisy, or missing heart rate readings. To tackle this issue, we augment the conventional Doppler ultrasound system with a rational agent that autonomously steers the ultrasound beam to track the position of the fetal heart. The proposed cognitive ultrasound system leverages a sequential Monte Carlo method to infer the fetal heart position from the power Doppler signal, and employs a greedy information-seeking criterion to select the steering angle that minimizes the positional uncertainty for future timesteps. The fetal heart rate is then calculated using the Doppler signal at the estimated fetal heart position. Our results show that the system can accurately track the fetal heart position across challenging signal-to-noise ratio scenarios, mainly thanks to its dynamic transmit beam steering capability. Additionally, we find that optimizing the transmit beamsteering to minimize positional uncertainty also optimizes downstream heart rate estimation performance. In conclusion, this work showcases the power of closed-loop cognitive ultrasound in boosting the capabilities of traditional systems.
This paper addresses the problem of accurate pose estimation-position, velocity, and orientation-of a rigid body using an Inertial Measurement Unit (IMU) in combination with generic exteroceptive measurements. By reformulating the vehicle's dynamics and measurement models within a linear time-varying (LTV) framework, we enable the application of a linear Kalman filter, significantly simplifying observer design for inertial navigation systems (INS). A key strength of this approach lies in its generality: rather than relying on specific measurement modalities, our framework accommodates a broad class of exteroceptive measurements. To illustrate its effectiveness, we conduct a uniform observability (UO) analysis for two fundamental benchmark cases-GPS-aided INS and landmark-aided INS-deriving sufficient conditions that guarantee the global uniform exponential stability of the proposed filter. Simulations for both applications confirm the versatility and robustness of our approach.
This paper addresses the problem of autonomous robot navigation in unknown, obstacle-filled environments with second-order dynamics by proposing a Dissipative Avoidance Feedback (DAF). Compared to the Artificial Potential Field (APF), which primarily uses repulsive forces based on position, DAF employs a dissipative feedback mechanism that accounts for both position and velocity, contributing to smoother and more natural obstacle avoidance. The proposed continuously differentiable controller solves the motion-to-goal problem while guaranteeing collision-free navigation by using the robot's state and local obstacle distance information. We show that the controller guarantees safe navigation in generic $n$-dimensional environments and that all undesired $\omega$-limit points are unstable under certain controlled curvature conditions. Designed for real-time implementation, DAF requires only locally measured data from limited-range sensors (e.g., LiDAR, depth cameras), making it particularly effective for robots navigating unknown workspaces. Simulations in 2D and 3D environments are conducted to validate the theoretical results and to showcase the effectiveness of our approach.
Multi-hop reasoning, which requires multi-step reasoning based on the supporting documents within a given context, remains challenging for large language models (LLMs). LLMs often struggle to filter out irrelevant documents within the context, and their performance is sensitive to the absolute position of supporting documents within that context. In this paper, we identify an additional challenge: LLMs' performance is also sensitive to the order, relative position, in which the supporting documents are presented. We refer to this as the misordered context problem. To address this issue, based on the theoretical approach, we propose a simple yet effective method called context repetition (CoRe), which involves prompting the model by repeatedly presenting the context. This ensures that certain contiguous reasoning segments within supporting documents are presented in the optimal order, effectively guiding the model's reasoning in the appropriate direction. Applying CoRe, we improve the F1 score by up to 30%p on multi-hop QA tasks and increase accuracy by up to 70%p on a synthetic task. Additionally, CoRe helps mitigate the well-known "lost-in-the-middle" problem in LLMs and can be effectively combined with retrieval-based approaches utilizing Chain-of-Thought (CoT) reasoning.
The Boolean Nearest Neighbor (BNN) representation of Boolean functions was recently introduced by Hajnal, Liu and Turan. A BNN representation of $f$ is a pair $(P,N)$ of sets of Boolean vectors (called positive and negative prototypes) where $f(x)=1$ for every positive prototype $x \in P$, $f(x)=0$ for all every negative prototype $x \in N$, and the value $f(x)$ for $x \not\in P \cup N$ is determined by the type of the closest prototype. The main aim of this paper is to determine the position of the BNN language in the Knowledge Compilation Map (KCM). To this end, we derive results which compare the succinctness of the BNN language to several standard languages from KCM, and determine the complexity status of most standard queries and transformations for BNN inputs.