Retrieval-Augmented Generation (RAG) faces a core bottleneck with knowledge-sparse and semantically ambiguous long-tail queries, where retrieval noise distorts reasoning and necessitates costly post-processing. To tackle this, we propose RaCoT (Retrieval-aware Contrastive-of-Thought), a novel framework that shifts contrastive thinking to the pre-retrieval stage. By automatically generating a semantically adjacent yet differently answered contrastive question and extracting a Δ-Prompt to capture their key differences, RaCoT guides the model to proactively focus on the "critical details that determine answer divergence." This approach allows it to suppress semantic interference within a single retrieval pass, overcoming the theoretical bottleneck of single-vector queries that struggle to simultaneously encode signals for what to attend to and what to ignore. On six authoritative benchmarks, including PopQA and TriviaQA-unfiltered, RaCoT outperforms strong baselines like RankRAG and Self-RAG by 0.9-2.4 percentage points. It exhibits superior robustness, with a performance drop of only 8.6% in adversarial tests, far surpassing the over 15% degradation in other methods. Furthermore, its low latency (3.12s) and token overhead (11.54) place it on the accuracy-efficiency Pareto frontier, while ablation studies validate the necessity of each component. Ultimately, RaCoT reframes the RAG paradigm from "post-hoc context cleaning" to "a priori shaping of discriminative reasoning," offering an efficient and robust path toward reliable AI systems for real-time, resource-constrained deployments.

Multi-agent systems (MAS) built on large language models (LLMs) often suffer from inefficient ''free-for-all'' communication, leading to exponential token costs and low signal-to-noise ratios that hinder their practical deployment. We challenge the notion that more communication is always beneficial, hypothesizing instead that the core issue is the absence of resource rationality. We argue that "free'' communication, by ignoring the principle of scarcity, inherently breeds inefficiency and unnecessary expenses. To address this, we introduce the Dynamic Auction-based Language Agent (DALA), a novel framework that treats communication bandwidth as a scarce and tradable resource. Specifically, our DALA regards inter-agent communication as a centralized auction, where agents learn to bid for the opportunity to speak based on the predicted value density of their messages. Thus, our DALA intrinsically encourages agents to produce concise, informative messages while filtering out low-value communication. Extensive and comprehensive experiments demonstrate that our economically-driven DALA achieves new state-of-the-art performance across seven challenging reasoning benchmarks, including 84.32% on MMLU and a 91.21% pass@1 rate on HumanEval. Note that this is accomplished with remarkable efficiency, i.e., our DALA uses only 6.25 million tokens, a fraction of the resources consumed by current state-of-the-art methods on GSM8K. Further analysis reveals that our DALA cultivates the emergent skill of strategic silence, effectively adapting its communication strategies from verbosity to silence in a dynamic manner via resource constraints.

Despite recent advancements in 3D-text cross-modal alignment, existing state-of-the-art methods still struggle to align fine-grained textual semantics with detailed geometric structures, and their alignment performance degrades significantly when scaling to large-scale 3D databases. To overcome this limitation, we introduce 3DAlign-DAER, a unified framework designed to align text and 3D geometry via the proposed dynamic attention policy and the efficient retrieval strategy, capturing subtle correspondences for diverse cross-modal retrieval and classification tasks. Specifically, during the training, our proposed dynamic attention policy (DAP) employs the Hierarchical Attention Fusion (HAF) module to represent the alignment as learnable fine-grained token-to-point attentions. To optimize these attentions across different tasks and geometric hierarchies, our DAP further exploits the Monte Carlo tree search to dynamically calibrate HAF attention weights via a hybrid reward signal and further enhances the alignment between textual descriptions and local 3D geometry. During the inference, our 3DAlign-DAER introduces an Efficient Retrieval Strategy (ERS) to leverage efficient hierarchical searching in the large-scale embedding spaces, outperforming traditional methods (eg, KNN) in accuracy and efficiency. Furthermore, to facilitate text-3D alignment research and train our 3DAlign-DAER, we construct Align3D-2M, a large-scale dataset featuring 2M text-3D pairs, to provide sufficient fine-grained cross-modal annotations. Extensive and comprehensive experiments demonstrate the superior performance of our 3DAlign-DAER on diverse benchmarks.

Large language models (LLMs) often generate hallucinated content lacking factual or contextual grounding, hindering their reliability in critical applications. Traditional methods like supervised fine-tuning and reinforcement learning from human feedback are data-intensive and computationally expensive, while static parameter editing struggles with context-dependent errors and catastrophic forgetting. To overcome these limitations, we introduce LLM-CAS, a framework that formulates real-time hallucination correction as a hierarchical reinforcement learning (HRL) problem. LLM-CAS trains an agent to learn a sophisticated policy, dynamically selecting optimal, temporary neuron perturbations during inference based on the immediate context. This learned, policy-driven approach provides greater adaptability than prior dynamic methods that rely on heuristic or pre-defined adjustments. As a result, LLM-CAS achieves significant performance gains across various LLMs, improving accuracy by 10.98 percentage points on StoryCloze, 2.71 points on TriviaQA, and 2.06 points on TruthfulQA's MC1 score, thereby outperforming static methods like ITI and CAA, as well as the dynamic SADI framework. This context-aware, efficient approach promises enhanced reliability for LLMs in high-stakes domains, with future potential for multimodal extensions.

Yijia Fan, Jing Yang, Mingyu Liu, Kaitong Cai, Jian Wang, Keze Wang, Jusheng Zhang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

This paper introduces DrDiff, a novel framework for long-text generation that overcomes the efficiency-quality trade-off through three core technologies.First, we design a dynamic expert scheduling mechanism that intelligently allocates computational resources during the diffusion process based on text complexity, enabling more efficient handling of text generation tasks of varying difficulty.Second, we introduce a Hierarchical Sparse Attention (HSA) mechanism that adaptively adjusts attention patterns according to a variety of input lengths, reducing computational complexity from O(n 2 ) to O(n) while maintaining model performance.Finally, we propose a Semantic Anchor States (SAS) module that combines with DPM-solver++ to reduce diffusion steps, significantly improving generation speed.Comprehensive experiments on various long-text generation benchmarks demonstrate the superiority of our DrDiff over the existing SOTA methods.

As scaling large language models faces prohibitive costs, multi-agent systems emerge as a promising alternative, though challenged by static knowledge assumptions and coordination inefficiencies. We introduce Knowledge-Aware Bayesian Bandits (KABB), a novel framework that enhances multi-agent system coordination through semantic understanding and dynamic adaptation. The framework features three key innovations: a customized knowledge distance model for deep semantic understanding, a dual-adaptation mechanism for continuous expert optimization, and a knowledge-aware Thompson Sampling strategy for efficient expert selection. Extensive evaluation demonstrates KABB achieves an optimal cost-performance balance, maintaining high performance while keeping computational demands relatively low in multi-agent coordination.

We propose a novel collaborative multi-agent optimization framework for adaptive training in multi-label image classification, fundamentally advancing beyond static decision rules and isolated automation. Our method deploys a set of distributed, task-specific agents, each responsible for dynamically orchestrating critical training components—including data augmentation, optimization methods, learning rate schedules, and loss functions—according to evolving visual-semantic relationships and training states. Each agent employs an advanced non-stationary multi-armed bandit algorithm, integrating both $\epsilon$-greedy and upper confidence bound strategies, to judiciously balance exploration with exploitation throughout the training lifecycle. A hierarchical composite reward mechanism synergizes overall classification accuracy, rare class recognition, and training stability, fostering both independent optimization and implicit collaborative behavior among agents. The framework further leverages refined techniques such as dual-rate exponential moving average smoothing and structured mixed-precision training to enhance robustness and computational efficiency. Extensive experiments across benchmarks including Pascal VOC, COCO, Yeast, and Mediamill demonstrate that our approach achieves superior mean average precision and rare-class F1 scores compared to state-of-the-art methods, while also exhibiting rapid convergence and remarkable cross-domain generalization. Our results indicate that collaborative multi-agent adaptive optimization offers a scalable and principled solution for self-optimizing deep learning in complex multi-label scenarios.

Recent advances in vision-language models (VLMs) have greatly improved cross-modal semantic understanding, yet significant limitations remain in fine-grained discrimination and deep causal reasoning tasks. Existing VLMs often rely on superficial statistical correlations, lacking the ability to capture the underlying causal logic between visual and textual content. To address this, we propose the **CounterFactual Vision-Language Fine-tuning Model (CF-VLM)**, a novel framework that enhances the causal reasoning capabilities of VLMs through the targeted use of counterfactual samples. CF-VLM introduces three complementary training objectives: maintaining foundational cross-modal alignment, reinforcing the uniqueness, and stability of factual scene representations against coherent counterfactuals, and sharpening the model’s sensitivity to minimal but critical causal edits. Extensive experiments demonstrate that CF-VLM consistently outperforms strong baselines and state-of-the-art methods on compositional reasoning and generalization benchmarks. Furthermore, it shows promise in mitigating visual hallucinations, indicating improved factual consistency. Our CF-VLM provides a robust foundation for deploying VLMs in high-stakes, real-world scenarios requiring reliable reasoning and interpretability.

We propose **GAM-Agent**, a game-theoretic multi-agent framework for enhancing vision-language reasoning. Unlike prior single-agent or monolithic models, GAM-Agent formulates the reasoning process as a non-zero-sum game between base agents—each specializing in visual perception subtasks—and a critical agent that verifies logic consistency and factual correctness. Agents communicate via structured claims, evidence, and uncertainty estimates. The framework introduces an uncertainty-aware controller to dynamically adjust agent collaboration, triggering multi-round debates when disagreement or ambiguity is detected. This process yields more robust and interpretable predictions. Experiments on four challenging benchmarks—MMMU, MMBench, MVBench, and V*Bench—demonstrate that GAM-Agent significantly improves performance across various VLM backbones. Notably, GAM-Agent boosts the accuracy of small-to-mid scale models (e.g., Qwen2.5-VL-7B, InternVL3-14B) by 5–6\%, and still enhances strong models like GPT-4o by up to 2–3\%. Our approach is modular, scalable, and generalizable, offering a path toward reliable and explainable multi-agent multimodal reasoning.

Driven by the applications in autonomous driving, robotics, and augmented reality, 3D object annotation is a critical task compared to 2D annotation, such as spatial complexity, occlusion, and viewpoint inconsistency. The existing methods relying on single models often struggle with these issues. In this paper, we introduce Tri-MARF, a novel framework that integrates tri-modal inputs (i.e., 2D multi-view images, text descriptions, and 3D point clouds) with multi-agent collaboration to enhance the 3D annotation process. Our Tri-MARF consists of three specialized agents: a vision-language model agent that generates multi-view descriptions, an information aggregation agent that selects optimal descriptions, and a gating agent that aligns text descriptions with 3D geometries for more refined captioning. Extensive experiments on the Objaverse-LVIS, Objaverse-XL, and ABO datasets demonstrate the superiority of our Tri-MARF, which achieves a CLIPScore of 88.7 (compared to 78.6–82.4 for other SOTA methods), retrieval accuracy of 45.2/43.8 (ViLT R@5), and an impressive throughput of 12,000 objects per hour on a single NVIDIA A100 GPU.

This paper studies the combined problem of base station location and optimal power allocation, in order to optimize the energy efficiency of a cellular wireless network. Recent work has suggested that moving from a network of a small number of high power macrocells to a larger number of smaller microcells may improve the energy efficiency of the network. This paper investigates techniques to optimize the number of base stations and their locations, in order to minimize energy consumption. An important contribution of the paper is that it takes into account non-uniform user distributions across the coverage area, which is likely to be encountered in practice. The problem is solved using approaches from optimization theory that deal with the facility location problem. Stochastic programming techniques are used to deal with the expected user distributions. An example scenario is presented to illustrate how the technique works and the potential performance gains that can be achieved.

A new precoder that reduces the feedback overhead in an OFDM-based relay network is proposed. A joint Singular Value Decomposition (SVD) and QR Decomposition is employed at the source-destination link and source-relay-destination link. Using the proposed precoder, only one precoding matrix needs to be fed back, independent of the total number of subcarriers. The link achievable rate using the reduced complexity precoder is evaluated via simulations and compared with that of full SVD precoders and for varying codebook size.

In this paper, a novel cooperative diversity transmission protocol is proposed for a four-node network where a single-antenna source communicates with its intended N-antenna destination with the help of two K-antenna decode-and-forward (DF) relays. Without requiring complex coding strategies at the relays, a sufficiently strong or weak inter-relay channel, or destination-source feedback, the proposed shifted successive DF relaying (SSDFR) protocol asymptotically achieves the optimal diversity-multiplexing tradeoff performance the four-node network can provide.

A novel cooperative diversity transmission protocol using two K-antenna decode-and-forward relays to take turns assisting in the communication between a single-antenna source and its intended N-antenna destination is studied. A (1+2⌈logK⌉)-bit destination-relay feedback signal is exploited to perform relay/antenna selection. Two different simple forwarding strategies are applied at the relays: one relay uses a superposition-coding strategy while the other relay uses a repetition-coding strategy. When the source's frame length is sufficiently large, the optimal diversity- multiplexing tradeoff of the four-node network can be achieved asymptotically if K ≥ 3.

In this letter the performance limits and design principles of rateless codes over fading channels are studied. The diversity-multiplexing tradeoff (DMT)is used to analyze the system performance for all possible transmission rates. It is revealed from the analysis that the design of such rateless codes follows the design principle of approximately universal codes for multiple-input multiple-output (MIMO) channels. It is also shown that for a single-input single-output (SISO) channel, simple permutation codes of unit length for parallel channels can be transformed directly into rateless codes that achieve the DMT performance limit of the channel.

A novel cooperative diversity transmission protocol is proposed for a four-node network, in which a single-antenna source communicates with an N-antenna destination with the help of two K-antenna half-duplex decode-and-forward (DF) relays. The proposed shifted successive DF relaying (SSDFR) protocol asymptotically achieves the optimal diversity-multiplexing tradeoff (DMT) the system can provide. The resulting DMT performance does not require complex coding strategies at the relays, extremely strong or weak inter-relay channel conditions, or feedback from the destination to the source, which highlights the advantages of the proposed protocol over state of the art techniques.

In this letter the performance of multiple relay channels is analyzed for the situation in which multiple antennas are deployed only at the relays. The simple repetition-coded decode and-forward protocol with two different antenna processing techniques at the relays is investigated. The antenna combining techniques are maximum ratio combining (MRC) for reception and transmit beamforming (TB) for transmission. It is shown that these distributed antenna combining techniques can exploit the full spatial diversity of the relay channels regardless of the number of relays and antennas at each relay, and offer significant power gain over distributed space-time coding techniques.

Previous work on cooperative communications has concentrated primarily on the diversity benefits of such techniques. This paper, instead, considers the multiplexing benefits of cooperative communications. First, a new interpretation on the fundamental tradeoff between the transmission rate and outage probability in multiple-antenna relay networks is given. It follows that multiplexing gains can be obtained at any finite signal-to-noise ratio (SNR), in full-duplex multiple-antenna relay networks. Thus, relaying can offer not only stronger link reliability, but also higher spectral efficiency. Specifically, the decode-and-forward protocol is applied and networks that have one source, one destination, and multiple relays are considered. A receive power gain at the relays, which captures the network large-scale fading characteristics, is also considered. It is shown that this power gain can significantly affect the system diversity-multiplexing tradeoff for any finite SNR value. Several relaying protocols are proposed and are shown to offer nearly the same outage probability as if the transmit antennas at the source and the relay(s) were colocated, given certain SNR and receive power gains at the relays. Thus, a higher multiplexing gain than that of the direct link can be obtained if the destination has more antennas than the source. Much of the analysis in the paper is valid for arbitrary channel fading statistics. These results point to a view of relay networks as a means for providing higher spectral efficiency, rather than only link reliability.

In this paper, a two-hop cooperative multi-relay communication network is considered. Selection relaying schemes are attracting considerable attention due to their prudent bandwidth utilization and ability to provide full diversity. The recent developments in selection relaying (SR) have largely focused on information theoretic analyses such as outage performance. Some of these analyses are accurate only in high SNR regimes. This paper provides exact outage and capacity performance expressions for selection relaying and tight approximation over a sufficiently wide range of SNR regimes for selection cooperative relaying. The outage capacity for SR is also provided. The motivation for this work is that practical systems operate at far lower SNR values than those supported by the high SNR analysis. Therefore, designers should be able to evaluate system performance to a reasonable degree of accuracy over practical SNR values. Simulations are used to corroborate the analytical results and close agreement is observed.

The diversity-multiplexing tradeoff (DMT) performances of novel decode-and-forward (DF) relaying protocols for multi-source multi-relay cooperative networks are studied in this paper. For a strong-interference scenario and an isolated-relay scenario, the proposed protocols significantly improve diversity performance over direct source-destination transmissions by using a simple repetition coding strategy in relays. This is in addition to enhancing the multiplexing performance over standard DF protocols, which usually suffer from the half-duplex limitation at relays. Although the DMT performance of DF based relaying protocols is limited by the quality of source relay links in principle, adaptive forms of the proposed protocols at the relays can transmit/not transmit according to source relay transmissions and still provide link reliability and spectral efficiency advantages in general source-relay channel conditions.

Consider a network with one source, one destination and multiple relays. It is shown that multiplexing gains that are higher than that of the direct link are possible for any finite SNR, in such a network, under any fading statistics. Thus relaying can offer not only stronger link reliability, but also higher spectral efficiency, compared with non-relay transmission. Several schemes for the decode-and-forward protocol are proposed and are shown to offer the same outage probability and diversity-multiplexing tradeoff (DMT) as if the transmit antennas at the source and the relay(s) were co-located, under certain power constraints. Thus a higher multiplexing gain than that of the direct link can be obtained if the destination has more antennas than the source.

In this work, the agility of detection algorithms in cognitive radio is studied. A sequential change detection framework is developed to investigate the delay of detection algorithms. Three scenarios with different information about primary users' parameters available at cognitive users are considered. Optimal detection schemes that minimize the detection delay under certain false alarm constraints are developed. Minimal detection delay is characterized as a function of the false alarm probability and the Kullback-Leibler distance between signal plus noise and noise only models.

This paper considers two-hop cooperative relaying using multiple relays and proposes a threshold based relay selection protocol. In this protocol the relays are selected among those having received SNR higher than a threshold value. The relay selection is performed by the destination based on the received SNRs at the destination during the last hop. The exact bit error rate of this protocol is derived and it is shown that it achieves full diversity order. Unlike some other full diversity achieving protocols in the literature, the requirement that the exact/average SNRs of the source-relay links be known at the destination is eliminated using an appropriate SNR threshold.

In this paper the performance limits and design principles of rateless codes over fading channels are studied. The diversity-multiplexing tradeoff (DMT) is used to analyze the system performance for all possible transmission rates. It is revealed from the analysis that the design of such rateless codes follows the design principle of approximately universal codes for parallel multiple-input multiple-output (MIMO) channels, in which each sub-channel is a MIMO channel. More specifically, it is shown that for a single-input single-output (SISO) channel, the previously developed permutation codes of unit length for parallel channels having rate LR can be transformed directly into rateless codes of length L having multiple rate levels (R, 2R, …, LR), to achieve the DMT performance limit.

In this paper, a superposition-coded concurrent decode-and-forward (DF) relaying protocol is presented. A specific scenario, where the inter-relay channel is sufficiently strong, is considered. Assuming perfect source-relay transmissions, the proposed scheme further improves the diversity performance of previously proposed repetition-coded concurrent DF relaying, in which the advantage of the inter-relay interference is not fully extracted.

Interest in selection relaying is growing. The recent developments in this area have largely focused on information theoretic analyses such as outage performance. Some of these analyses are accurate only at high SNR regimes. In this paper error rate analyses that are sufficiently accurate over a wide range of SNR regimes are provided. The motivations for this work are that practical systems operate at far lower SNR values than those supported by the high SNR analysis. To enable designers to make informed decisions regarding network design and deployment, it is imperative that system performance is evaluated with a reasonable degree of accuracy over practical SNR regimes. Simulations have been used to corroborate the analytical results, as close agreement between the two is observed.

In uncoded cooperative wireless networks, error propagation due to the detection errors at the relay limits the performance of cooperative digital relaying. Threshold relaying is a simple and effective technique to mitigate error propagation without any reliance on the channel coding. This paper analyzes the asymptotic end-to-end (e2e) bit error rate (BER) of threshold digital relaying in a network with a single relay and links experiencing independent Rayleigh fading. It is shown that the optimal threshold that minimizes the e2e BER increases as log(SNR) as the average link signal-to-noise ratios are increased simultaneously. The resulting e2e BER decreases as log(SNR)/SNR2, thereby achieving dual diversity.

In this paper, the diversity-multiplexing tradeoff (DMT) behavior of a novel concurrent decode-and-forward (DF) relaying cooperative diversity transmission protocol is analyzed. A two-source two-relay one-destination scenario is considered in which concurrent transmission among the network nodes is used by combining the two sources' two classic DF relaying steps. Through the DMT analysis, it is shown that the proposed protocol can effectively recover the multiplexing loss induced by the classic DF relaying protocol, while still obtaining diversity gain. The system model is further extended to a generalized M-source network.

This paper studies selective relaying schemes based on signal-to-noise-ratio (SNR) to minimize the end-to-end (e2e) bit error rate (BER) in cooperative digital relaying systems using BPSK modulation. In the SNR-based selective relaying, the relay either retransmits or remains silent depending on the SNRs of the source-relay, relay-destination, and source-destination links. Different models assuming the availability of different sets of instantaneous and average SNR information at the relay are studied. For each model, the optimal strategy to minimize the e2e BER is a different threshold rule on the source-relay SNR, if the link SNRs are uncorrelated in time and space. Approximations for the optimal threshold values that minimize the e2e BER and the resulting performance are derived analytically for BPSK modulation. Using the derived threshold the e2e BER can be reduced significantly compared to simple digital relaying. By studying the performance under different models, it is shown that knowledge of the instantaneous source-destination SNR at the relay can be exploited. The gain from this knowledge is higher when the average source-destination SNR is large. However, knowledge of the instantaneous relay-destination SNR at the relay does not change performance significantly.

Threshold relaying is an effective technique to achieve cooperative diversity in uncoded cooperative wireless networks, which suffer from error propagation due to detection errors at the relays. This paper analyzes the asymptotic end-to-end (e2e) bit error rate (BER) of threshold digital relaying. A three node network with a source, destination and relay and with links experiencing independent Rayleigh fading is considered. It is shown that, as the average link signal-to-noise ratios (SNR) are increased simultaneously, the optimal threshold that minimizes the e2e BER increases as log(SNR). The resulting e2e BER decreases as log(SNR)/SNR2. Moreover, any threshold of the form log(cSNR), where c is a positive constant, achieves the same order of e2e BER as the one achieved by the optimal threshold and provides dual diversity. A value of c that performs very close to the optimal threshold is also proposed.

This paper explores two buffering relay models to improve the capacity of relay networks in slow fading environments. Throughput, average packet delay, and information loss are considered as functions of buffer size and signal-to-noise ratio (SNR). It is shown that, for any fading statistics, both a fixed buffering relay model and a dynamic buffering relay model offer significant performance advantages in terms of capacity over existing methods at the expense of increased delay. It is also seen that, of the two models, dynamic buffering provides a smaller average delay.

Recent theoretical and simulation studies reveal that closely coupled antennas with appropriately chosen impedance matching loads can yield the desired characteristic of high MIMO capacity. Because of the practical infeasibility of multiport- conjugate matching, the single-port matching impedance (Zopt) which maximizes the capacity of a 2 times 2 MIMO system with coupled half-wavelength dipoles is derived in this paper. We derive Zoptfor an ergodic capacity upper bound including the mutual coupling and matching networks effect. We show that the closed-form of Zoptin the high SNR regime is identical to the input impedance. The analytical results are validated by numerical simulations.

In this paper we discuss and compare different signalling and relaying methods for multiple-input multiple-output (MIMO) relay networks in terms of network capacity, where every terminal is equipped with multiple antennas. We propose a new relaying mode called hybrid relaying, by which the MIMO relay channels can be decomposed into several point-to-point parallel channels without decoding the desired signals at the relay. We show that the proposed hybrid relaying algorithms outperform the conventional analogue relaying scheme. They can be good suboptimal choices compared with digital relying schemes, providing an attractive tradeoff between performance and complexity, especially when larger numbers of antennas are deployed at the relay than at the source and destination.

In this paper we analyze the performance of multiple relay channels when multiple antennas are deployed only at relays. Specifically, we investigate the simple repetition-coded decode-and-forward protocol and apply two antenna combining techniques at relays, namely maximum ratio combining (MRC) on receive and transmit beamforming (TB). We assume that the total number of antennas at all relays is fixed to N. With a reasonable power constraint at the relays, we show that the antenna combining techniques can exploit the full spatial diversity of the relay channels and can achieve the same diversity multiplexing tradeoff as achieved by more complex space-time distributed coding techniques, such as those proposed by Laneman and Womell (2003).

We study selective digital relaying schemes where the relay may choose to retransmit or to remain silent based on the qualities of the links between the source, relay and the destination. We first analyze a baseline scheme, called static relaying, where the relaying decisions are based only on the average signal-to-noise ratio (SNR) values of all the links. The second scheme, dynamic relaying, allows the relay to make decisions based on the instantaneous SNR of the source-relay link and average SNRs of the relay-destination and source-destination links. We show that, in dynamic relaying the optimal strategy to minimize the average end-to-end bit error rate is a threshold rule on the instantaneous SNR of the source-relay channel. In this case, the optimal threshold value is a function of average SNR of relay-destination and source-destination channels. We derive closed-form expressions for the optimal threshold and the bit error performance achieved by this threshold. We show that dynamic relaying can provide significant performance advantage over static relaying.

In this paper we discuss and compare different signalling and routing methods for multiple-input multiple-output (MIMO) relay networks in terms of the network capacity, where every terminal is equipped with multiple antennas. Our study for signalling includes the two well known digital (decode and forward) relaying, analogue (amplify and forward) relaying, and a novel hybrid (filter, amplify and forward) relaying. We propose both optimal and suboptimal hybrid relaying schemes which avoid full decoding of the message at the relay. We show that they outperform analogue relaying and give similar performance to digital relaying, particularly when the relay has forward channel state information (CSI) or larger number of antennas than the source and destination. For the routing schemes designed for multiple relay channels, we use relay selection schemes to exploit the spatial diversity, which we call selection diversity of the networks. We propose both optimal and suboptimal relay selection schemes and show that their performance converges when a large number of antennas is deployed at each node in the network. We also compare relay selection routing with a space-time coded relay cooperation protocol and show the performance advantage of selection diversity over cooperative diversity in certain scenarios. Finally, we give a brief discussion on the application of another MIMO structure called single signal beamforming in the relay scenario. Its performance will be compared with that of spatial multiplexing

In this paper, a transmission protocol is studied for a two relay wireless network in which simple repetition coding is applied at the relays. Information-theoretic achievable rates for this transmission scheme are given, and a space-time V-BLAST signalling and detection method that can approach them is developed. It is shown through the diversity multiplexing tradeoff analysis that this transmission scheme can recover the multiplexing loss of the half-duplex relay network, while retaining some diversity gain. This scheme is also compared with conventional transmission protocols that exploit only the diversity of the network at the cost of a multiplexing loss. It is shown that the new transmission protocol offers significant performance advantages over conventional protocols, especially when the interference between the two relays is sufficiently strong.

In this paper we discuss the outage capacity of multihop networks employing terminals with multiple antennas. We discuss the performance of several novel relaying configurations and signalling algorithms for either limited feedback or non-feedback channels. Three relaying types are considered, namely analogue relaying, digital relaying and hybrid relaying. We find that digital selective relaying outperforms all other relaying configurations considered here while requiring the highest signalling overhead. The matched filter based hybrid relaying scheme appears to be a good suboptimum choice for its good performance and low signalling overhead in the multiple relay scenario. We also find that for hybrid relaying schemes, the MMSE algorithm, which is usually applied in conventional MIMO systems, might not give effective performance improvement when it is applied at relays.