Search

Large Language Models (LLMs) have the potential to revolutionize the Sixth Generation (6G) communication networks. However, current mainstream LLMs generally lack the specialized knowledge in telecom domain. In this paper, for the first time, we propose a pipeline to adapt any general purpose LLMs to a telecom-specific LLMs. We collect and build telecom-specific pre-train dataset, instruction dataset, preference dataset to perform continual pre-training, instruct tuning and alignment tuning respectively. Besides, due to the lack of widely accepted evaluation benchmarks in telecom domain, we extend existing evaluation benchmarks and proposed three new benchmarks, namely, Telecom Math Modeling, Telecom Open QnA and Telecom Code Tasks. These new benchmarks provide a holistic evaluation of the capabilities of LLMs including math modeling, Open-Ended question answering, code generation, infilling, summarization and analysis in telecom domain. Our fine-tuned LLM TelecomGPT outperforms state of the art (SOTA) LLMs including GPT-4, Llama-3 and Mistral in Telecom Math Modeling benchmark significantly and achieve comparable performance in various evaluation benchmarks such as TeleQnA, 3GPP technical documents classification, telecom code summary and generation and infilling., Comment: arXiv admin note: text overlap with arXiv:1303.2654 by other authors

Integrated sensing and communication (ISAC) has emerged as a promising technology to facilitate high-rate communications and super-resolution sensing, particularly operating in the millimeter wave (mmWave) band. However, the vulnerability of mmWave signals to blockages severely impairs ISAC capabilities and coverage. To tackle this, an efficient and low-cost solution is to deploy distributed reconfigurable intelligent surfaces (RISs) to construct virtual links between the base stations (BSs) and users in a controllable fashion. In this paper, we investigate the generalized RIS-assisted mmWave ISAC networks considering the blockage effect, and examine the beneficial impact of RISs on the coverage rate utilizing stochastic geometry. Specifically, taking into account the coupling effect of ISAC dual functions within the same network topology, we derive the conditional coverage probability of ISAC performance for two association cases, based on the proposed beam pattern model and user association policies. Then, the marginal coverage rate is calculated by combining these two cases through the distance-dependent thinning method. Simulation results verify the accuracy of derived theoretical formulations and provide valuable guidelines for the practical network deployment. Specifically, our results indicate the superiority of the RIS deployment with the density of 40 km${}^{-2}$ BSs, and that the joint coverage rate of ISAC performance exhibits potential growth from $67.1\%$ to $92.2\%$ with the deployment of RISs.

Millimeter-wave (mmWave) multiple-input multiple-output (MIMO) communication with the advanced beamforming technologies is a key enabler to meet the growing demands of future mobile communication. However, the dynamic nature of cellular channels in large-scale urban mmWave MIMO communication scenarios brings substantial challenges, particularly in terms of complexity and robustness. To address these issues, we propose a robust gradient-based liquid neural network (GLNN) framework that utilizes ordinary differential equation-based liquid neurons to solve the beamforming problem. Specifically, our proposed GLNN framework takes gradients of the optimization objective function as inputs to extract the high-order channel feature information, and then introduces a residual connection to mitigate the training burden. Furthermore, we use the manifold learning technique to compress the search space of the beamforming problem. These designs enable the GLNN to effectively maintain low complexity while ensuring strong robustness to noisy and highly dynamic channels. Extensive simulation results demonstrate that the GLNN can achieve 4.15% higher spectral efficiency than that of typical iterative algorithms, and reduce the time consumption to only 1.61% that of conventional methods., Comment: Accepted by IEEE Wireless Communications Letters

Generative artificial intelligence (GenAI) and communication networks are expected to have groundbreaking synergies in 6G. Connecting GenAI agents over a wireless network can potentially unleash the power of collective intelligence and pave the way for artificial general intelligence (AGI). However, current wireless networks are designed as a "data pipe" and are not suited to accommodate and leverage the power of GenAI. In this paper, we propose the GenAINet framework in which distributed GenAI agents communicate knowledge (high-level concepts or abstracts) to accomplish arbitrary tasks. We first provide a network architecture integrating GenAI capabilities to manage both network protocols and applications. Building on this, we investigate effective communication and reasoning problems by proposing a semantic-native GenAINet. Specifically, GenAI agents extract semantic concepts from multi-modal raw data, build a knowledgebase representing their semantic relations, which is retrieved by GenAI models for planning and reasoning. Under this paradigm, an agent can learn fast from other agents' experience for making better decisions with efficient communications. Furthermore, we conduct two case studies where in wireless device query, we show that extracting and transferring knowledge can improve query accuracy with reduced communication; and in wireless power control, we show that distributed agents can improve decisions via collaborative reasoning. Finally, we address that developing a hierarchical semantic level Telecom world model is a key path towards network of collective intelligence.

In an era marked by unprecedented climatic upheavals and evolving urban landscapes, the role of advanced communication networks in disaster prevention and management is becoming increasingly critical. This paper explores the transformative potential of 3D wireless networks, an innovative amalgamation of terrestrial, aerial, and satellite technologies, in enhancing disaster response mechanisms. We delve into a myriad of use cases, ranging from large facility evacuations to wildfire management, underscoring the versatility of these networks in ensuring timely communication, real-time situational awareness, and efficient resource allocation during crises. We also present an overview of cutting-edge prototypes, highlighting the practical feasibility and operational efficacy of 3D wireless networks in real-world scenarios. Simultaneously, we acknowledge the challenges posed by aspects such as cybersecurity, cross-border coordination, and physical layer technological hurdles, and propose future directions for research and development in this domain.

In this paper, we study the problem of joint active and passive beamforming for reconfigurable intelligent surface (RIS)-assisted massive multiple-input multiple-output systems towards the extension of the wireless cellular coverage in 3D, where multiple RISs, each equipped with an array of passive elements, are deployed to assist a base station (BS) to simultaneously serve multiple unmanned aerial vehicles (UAVs) in the same time-frequency resource of 5G wireless communications. With a focus on ensuring fairness among UAVs, our objective is to maximize the minimum signal-to-interference-plus-noise ratio (SINR) at UAVs by jointly optimizing the transmit beamforming parameters at the BS and phase shift parameters at RISs. We propose two novel algorithms to address this problem. The first algorithm aims to mitigate interference by calculating the BS beamforming matrix through matrix inverse operations once the phase shift parameters are determined. The second one is based on the principle that one RIS element only serves one UAV and the phase shift parameter of this RIS element is optimally designed to compensate the phase offset caused by the propagation and fading. To obtain the optimal parameters, we utilize one state-of-the-art reinforcement learning algorithm, deep deterministic policy gradient, to solve these two optimization problems. Simulation results are provided to illustrate the effectiveness of our proposed solution and some insightful remarks are observed.

The emerging concept of extremely-large holographic multiple-input multiple-output (HMIMO), beneficial from compactly and densely packed cost-efficient radiating meta-atoms, has been demonstrated for enhanced degrees of freedom even in pure line-of-sight conditions, enabling tremendous multiplexing gain for the next-generation communication systems. Most of the reported works focus on energy and spectrum efficiency, path loss analyses, and channel modeling. The extension to secure communications remains unexplored. In this paper, we theoretically characterize the secrecy capacity of the HMIMO network with multiple legitimate users and one eavesdropper while taking into consideration artificial noise and max-min fairness. We formulate the power allocation (PA) problem and address it by following successive convex approximation and Taylor expansion. We further study the effect of fixed PA coefficients, imperfect channel state information, inter-element spacing, and the number of Eve's antennas on the sum secrecy rate. Simulation results show that significant performance gain with more than 100\% increment in the high signal-to-noise ratio (SNR) regime for the two-user case is obtained by exploiting adaptive/flexible PA compared to the case with fixed PA coefficients., Comment: 7 pages, 7 figures, submitted to IEEE ICC 2024

Wireless communications at high-frequency bands with large antenna arrays face challenges in beam management, which can potentially be improved by multimodality sensing information from cameras, LiDAR, radar, and GPS. In this paper, we present a multimodal transformer deep learning framework for sensing-assisted beam prediction. We employ a convolutional neural network to extract the features from a sequence of images, point clouds, and radar raw data sampled over time. At each convolutional layer, we use transformer encoders to learn the hidden relations between feature tokens from different modalities and time instances over abstraction space and produce encoded vectors for the next-level feature extraction. We train the model on a combination of different modalities with supervised learning. We try to enhance the model over imbalanced data by utilizing focal loss and exponential moving average. We also evaluate data processing and augmentation techniques such as image enhancement, segmentation, background filtering, multimodal data flipping, radar signal transformation, and GPS angle calibration. Experimental results show that our solution trained on image and GPS data produces the best distance-based accuracy of predicted beams at 78.44%, with effective generalization to unseen day scenarios near 73% and night scenarios over 84%. This outperforms using other modalities and arbitrary data processing techniques, which demonstrates the effectiveness of transformers with feature fusion in performing radio beam prediction from images and GPS. Furthermore, our solution could be pretrained from large sequences of multimodality wireless data, on fine-tuning for multiple downstream radio network tasks.

In this work, we study the problem of semantic communication and inference, in which a student agent (i.e. mobile device) queries a teacher agent (i.e. cloud sever) to generate higher-order data semantics living in a simplicial complex. Specifically, the teacher first maps its data into a k-order simplicial complex and learns its high-order correlations. For effective communication and inference, the teacher seeks minimally sufficient and invariant semantic structures prior to conveying information. These minimal simplicial structures are found via judiciously removing simplices selected by the Hodge Laplacians without compromising the inference query accuracy. Subsequently, the student locally runs its own set of queries based on a masked simplicial convolutional autoencoder (SCAE) leveraging both local and remote teacher's knowledge. Numerical results corroborate the effectiveness of the proposed approach in terms of improving inference query accuracy under different channel conditions and simplicial structures. Experiments on a coauthorship dataset show that removing simplices by ranking the Laplacian values yields a 85% reduction in payload size without sacrificing accuracy. Joint semantic communication and inference by masked SCAE improves query accuracy by 25% compared to local student based query and 15% compared to remote teacher based query. Finally, incorporating channel semantics is shown to effectively improve inference accuracy, notably at low SNR values.

The evolution of generative artificial intelligence (GenAI) constitutes a turning point in reshaping the future of technology in different aspects. Wireless networks in particular, with the blooming of self-evolving networks, represent a rich field for exploiting GenAI and reaping several benefits that can fundamentally change the way how wireless networks are designed and operated nowadays. To be specific, large GenAI models are envisioned to open up a new era of autonomous wireless networks, in which multi-modal GenAI models trained over various Telecom data, can be fine-tuned to perform several downstream tasks, eliminating the need for building and training dedicated AI models for each specific task and paving the way for the realization of artificial general intelligence (AGI)-empowered wireless networks. In this article, we aim to unfold the opportunities that can be reaped from integrating large GenAI models into the Telecom domain. In particular, we first highlight the applications of large GenAI models in future wireless networks, defining potential use-cases and revealing insights on the associated theoretical and practical challenges. Furthermore, we unveil how 6G can open up new opportunities through connecting multiple on-device large GenAI models, and hence, paves the way to the collective intelligence paradigm. Finally, we put a forward-looking vision on how large GenAI models will be the key to realize self-evolving networks.

The recent progress of artificial intelligence (AI) opens up new frontiers in the possibility of automating many tasks involved in Telecom networks design, implementation, and deployment. This has been further pushed forward with the evolution of generative artificial intelligence (AI), including the emergence of large language models (LLMs), which is believed to be the cornerstone toward realizing self-governed, interactive AI agents. Motivated by this, in this paper, we aim to adapt the paradigm of LLMs to the Telecom domain. In particular, we fine-tune several LLMs including BERT, distilled BERT, RoBERTa and GPT-2, to the Telecom domain languages, and demonstrate a use case for identifying the 3rd Generation Partnership Project (3GPP) standard working groups. We consider training the selected models on 3GPP technical documents (Tdoc) pertinent to years 2009-2019 and predict the Tdoc categories in years 2020-2023. The results demonstrate that fine-tuning BERT and RoBERTa model achieves 84.6% accuracy, while GPT-2 model achieves 83% in identifying 3GPP working groups. The distilled BERT model with around 50% less parameters achieves similar performance as others. This corroborates that fine-tuning pretrained LLM can effectively identify the categories of Telecom language. The developed framework shows a stepping stone towards realizing intent-driven and self-evolving wireless networks from Telecom languages, and paves the way for the implementation of generative AI in the Telecom domain.

Wireless communication in millimetre wave bands, namely above 20 GHz and up to 300 GHz, is foreseen as a key enabler technology for the next generation of wireless systems. The huge available bandwidth is contemplated to achieve high data-rate wireless communications, and hence, to fulfil the requirements of future wireless networks. In this paper, we discuss and illustrate new paradigms for the sub-THz physical layer, which either aim at maximizing the spectral efficiency, minimizing the device complexity, or finding good tradeoff. The solutions offered by appropriate modulation schemes and multi-antenna systems are assessed based on various potential scenarios.

Many algorithms for visible light positioning (VLP) localization do not consider the shapes of the transmitters, which leads to the impracticality of the algorithm and the low localization accuracy. Therefore, this paper proposes a novel VLP algorithm and addresses the problems in terms of practicality and complexity by using one non-point transmitter based on a monocular. Because the shape of the transmitter is considered, the proposed algorithm is easy for practice and has wide applicability. Besides, it decreases the computation for simple geometric model and expands the coverage for there is a greater chance of receiving signals from one light than that of receiving signals from multiple lights.

To meet the exponentially increasing air traffic, L-band (960-1164 MHz) digital aeronautical communication system (LDACS) has been introduced. The LDACS aims to exploit the vacant spectrum between incumbent Distance Measuring Equipment (DME) signals and envisioned to follow an orthogonal frequency division multiplexing (OFDM) approach to support high-speed delay-sensitive multimedia services. This paper deals with the design and implementation of end-to-end LDACS the transceiver on the Zynq System on Chip platform, consisting of FPGA as programmable logic (PL) and ARM as processing system (PS). We consider OFDM based LDACS and improve it further using windowing and/or filtering. We propose a hardware-software co-design approach and analyze various transceiver configurations by dividing it into PL and PS. We demonstrate the flexibility offered by such a co-design approach to choose the configuration as well as word-length for a given area, delay and power constraints. The transceiver is also integrated with the programmable analog front-end to validate its functionality in the presence of various RF impairments and wireless channels and interference specific to the LDACS environment. Via in-depth performance analysis concerning parameters such as out-of-band attenuation, DME interference, bit-error-rate, word-length, and complexity, we demonstrate wide bandwidth filtered OFDM as an attractive solution for the next generation LDACS.

Future 5G networks will serve a variety of applications that will coexist on the same spectral band and geographical area, in an uncoordinated and asynchronous manner. It is widely accepted that using CP-OFDM, the waveform used by most current communication systems, will make it difficult to achieve this paradigm. Especially, CP-OFDM is not adapted for spectral coexistence because of its poor spectral localization. Therefore, it has been widely suggested to use filter bank based multi carrier (FB-MC) waveforms with enhanced spectral localization to replace CP-OFDM. Especially, FB-MC waveforms are expected to facilitate coexistence with legacy CP-OFDM based systems. However, this idea is based on the observation of the PSD of FB-MC waveforms only. In this paper, we demonstrate that this approach is flawed and show what metric should be used to rate interference between FB-MC and CP-OFDM systems. Finally, our results show that using FB-MC waveforms does not facilitate coexistence with CP-OFDM based systems to a high extent., Comment: Manuscript submitted for review to IEEE Transactions on Wireless Communications

A coexistence scenario between filter bank (FB) based waveforms and legacy Cyclic Prefix-Orthogonal Frequency Division Multiplexing (CP-OFDM) is studied. It is shown that studies available on that matter use a wrong metric to rate the interference between the coexisting systems. Specifically, it is shown that even well spectrally localized FB waveforms interfere at a high level with incumbent CP-OFDM receivers. The presented results invalidate a number of studies in the literature, which stated that FB based waveforms could be very efficiently used to insert communications in parts of spectrum left free by incumbent legacy CP-OFDM users., Comment: 5 pages, 6 figures, Invited paper at ISWCS 2016 Special Session on Non-Orthogonal Access

To optimize the use of the spectrum, it is expected that the next generation of wireless networks (5G) will enable coexistence of newly introduced services with legacy cellular networks. These new services, like Device-To-Device (D2D) communication, should require limited synchronization with the legacy cell to limit the amount of signaling overhead in the network. However, it is known that Cyclic Prefix-Orthogonal Frequency Division Multiplexing (CP-OFDM) used in Long Term Evolution-Advanced (LTE-A) is not fit for asynchronous environments. This has motivated the search for a new waveform, able to enhance coexistence with CP-OFDM. Namely, it has been widely suggested that new devices could use OFDM/Offset-Quadrature Amplitude Modulation (OFDM/OQAM) to reduce the interference they inject to legacy cellular users. However, values of interference are usually measured at the input antenna of the receiver, based on the PSD of the interfering signal. We showed in previous works that this measurement is not representative of the actual interference that is seen after the demodulation operations. Building on this finding, we provide in this paper the first exact closed forms of cross-interference between OFDM/OQAM and CP-OFDM users. Our results prove that using OFDM/OQAM only marginally reduces interference to legacy users, in contradiction with many results in the literature., Comment: 5 pages, 7 figures, Accepted at IEEE Globecom Workshops 2016

Device-to-device (D2D) communication is being heralded as an important part of the solution to the capacity problem in future networks, and is expected to be natively supported in 5G. Given the high network complexity and required signalling overhead associated with achieving synchronization in D2D networks, it is necessary to study asynchronous D2D communications. In this paper, we consider a scenario whereby asynchronous D2D communication underlays an OFDMA macro-cell in the uplink. Motivated by the superior performance of new waveforms with increased spectral localization in the presence of frequency and time misalignments, we compare the system-level performance of a set-up for when D2D pairs use either OFDM or FBMC/OQAM. We first demonstrate that inter-D2D interference, resulting from misaligned communications, plays a significant role in clustered D2D topologies. We then demonstrate that the resource allocation procedure can be simplified when D2D pairs use FBMC/OQAM, since the high spectral localization of FBMC/OQAM results in negligible inter-D2D interference. Specifically, we identify that FBMC/OQAM is best suited to scenarios consisting of small, densely populated D2D clusters located near the encompassing cell's edge., Comment: 7 pages, 9 figures, Accepted at IEEE Globecom 2016 Workshops

To answer the challenges put out by the next generation of wireless networks (5G), important research efforts have been undertaken during the last few years to find new waveforms that are better spectrally localized and less sensitive to asynchronism effects than the widely deployed Cyclic Prefix Orthogonal Frequency Division Multiplexing (CP-OFDM). One of the most studied schemes is OFDM-Offset Quadrature Amplitude Modulation (OFDM/OQAM) based on the PHYDYAS filter pulse. In the recent literature, spectrum coexistence between OFDM/OQAM and CP-OFDM is commonly studied based on the Power Spectral Density (PSD) model. In this paper, we show that this approach is flawed and we show that the actual interference injected by OFDM/OQAM systems onto CP-OFDM is much higher than what is classically expected with the PSD based model in the literature. We show that though using OFDM/OQAM in secondary systems is still advantageous, it brings limited gain in the context of coexistence with incumbent CP-OFDM systems., Comment: 7 pages, 9 figures, ICT 2016

This paper analyses a scenario where a Device-To-Device (D2D) pair coexists with an Orthogonal Frequency Division Multiplexing (OFDM) based incumbent network. D2D transmitter communicates in parts of spectrum left free by cellular users, while respecting a given spectral mask. The D2D pair is misaligned in time and frequency with the cellular users. Furthermore, the D2D pair utilizes alternative waveforms to OFDM proposed for 5G. In this study, we show that it is not worth synchronising the D2D pair in time with respect to the cellular users. Indeed, the interference injected into the incumbent network has small variations with respect to time misalignment. We provide interference tables that encompass both time and frequency misalignment. We use them to analyse the maximum rate achievable by the D2D pair when it uses different waveforms. Then, we present numerical results showing what waveform should be utilized by the D2D pair according to the time-frequency resources that are not used by the incumbent network. Our results show that the delay induced by linearly convolved waveforms make them hardly applicable to short time windows, but that they dominate OFDM for long transmissions, mainly in the case where cellular users are very sensitive to interference., Comment: 7 pages, 7 figures, Accepted at IEEE ICC 2016

In this letter, the problem of optimal resource power allocation and relay selection for two way relaying cognitive radio networks using half duplex Decode and Forward (DF) and Amplify and Forward (AF) systems are investigated. The primary and secondary networks are assumed to access the spectrum at the same time, so that the interference introduced to the primary network caused by the secondary network should be below a certain interference threshold. In addition, a selection strategy between the AF and DF schemes is applied depending on the achieved secondary sum rate without affecting the quality of service of the primary network. A suboptimal approach based on a genetic algorithm is also presented to solve our problem. Selected simulation results show that the proposed suboptimal algorithm offers a performance close to the performance of the optimal solution with a considerable complexity saving.