Reconfigurable intelligent surface (RIS) is one of the most promising technologies for 5G-Adv and 6G. It has the characteristics of low cost, low complexity, and easy deployment, which provides a new opportunity to develop intelligent high-speed rail
way communications. The typical applications of RIS-assisted smart high-speed railway communications are introduced in detail, including suppressing the Doppler shift effect, solving frequent handoff problems, overcoming high penetration loss problems, and supporting high-precision train positioning. The key technologies of RIS-assisted smart high-speed railway communications are discussed in-depth, including channel measurement and modeling, channel estimation and feedback, beamforming, network architecture, and network deployment. It is believed that the combination of the new intelligent high-speed railway infrastructure and the new electromagnetic infrastructure built by RIS will bring broad industrial prospects to the intelligent high-speed railway in the future.
Cell-free (CF) massive multiple-input multiple-output (MIMO) is a promising solution to provide uniform good performance for unmanned aerial vehicle (UAV) communications. In this paper, we propose the UAV communication with wireless power transfer (W
PT) aided CF massive MIMO systems, where the harvested energy (HE) from the downlink WPT is used to support both uplink data and pilot transmission. We derive novel closed-form downlink HE and uplink spectral efficiency (SE) expressions that take hardware impairments of UAV into account. UAV communications with current small cell (SC) and cellular massive MIMO enabled WPT systems are also considered for comparison. It is significant to show that CF massive MIMO achieves two and five times higher 95%-likely uplink SE than the ones of SC and cellular massive MIMO, respectively. Besides, the large-scale fading decoding receiver cooperation can reduce the interference of the terrestrial user. Moreover, the maximum SE can be achieved by changing the time-splitting fraction. We prove that the optimal time-splitting fraction for maximum SE is determined by the number of antennas, altitude and hardware quality factor of UAVs. Furthermore, we propose three UAV trajectory design schemes to improve the SE. It is interesting that the angle search scheme performs best than both AP search and line path schemes. Finally, simulation results are presented to validate the accuracy of our expressions.
In this paper, we investigate the impact of channel aging on the performance of cell-free (CF) massive multiple-input multiple-output (MIMO) systems with both spatial correlation and pilot contamination. We derive novel closed-form uplink and downlin
k spectral efficiency (SE) expressions that take imperfect channel estimation into account. More specifically, we consider large-scale fading decoding and matched-filter receiver cooperation in the uplink. The uplink performance of a small-cell (SC) system is derived for comparison. The CF massive MIMO system achieves higher 95%-likely uplink SE than the SC system. In the downlink, the coherent transmission has four times higher 95%-likely per-user SE than the non-coherent transmission. Statistical channel cooperation power control (SCCPC) is used to mitigate the inter-user interference. SCCPC performs better than full power transmission, but the benefits are gradually weakened as the channel aging becomes stronger. Furthermore, strong spatial correlation reduces the SE but degrades the effect of channel aging. Increasing the number of antennas can improve the SE while decreasing the energy efficiency. Finally, we use the maximum normalized Doppler shift to design the SE-improved length of the resource block. Simulation results are presented to validate the accuracy of our expressions and prove that the CF massive MIMO system is more robust to channel aging than the SC system.
A simile is a figure of speech that directly makes a comparison, showing similarities between two different things, e.g. Reading papers can be dull sometimes,like watching grass grow. Human writers often interpolate appropriate similes into proper lo
cations of the plain text to vivify their writings. However, none of existing work has explored neural simile interpolation, including both locating and generation. In this paper, we propose a new task of Writing Polishment with Simile (WPS) to investigate whether machines are able to polish texts with similes as we human do. Accordingly, we design a two-staged Locate&Gen model based on transformer architecture. Our model firstly locates where the simile interpolation should happen, and then generates a location-specific simile. We also release a large-scale Chinese Simile (CS) dataset containing 5 million similes with context. The experimental results demonstrate the feasibility of WPS task and shed light on the future research directions towards better automatic text polishment.
In multi-turn dialog, utterances do not always take the full form of sentences cite{Carbonell1983DiscoursePA}, which naturally makes understanding the dialog context more difficult. However, it is essential to fully grasp the dialog context to genera
te a reasonable response. Hence, in this paper, we propose to improve the response generation performance by examining the models ability to answer a reading comprehension question, where the question is focused on the omitted information in the dialog. Enlightened by the multi-task learning scheme, we propose a joint framework that unifies these two tasks, sharing the same encoder to extract the common and task-invariant features with different decoders to learn task-specific features. To better fusing information from the question and the dialog history in the encoding part, we propose to augment the Transformer architecture with a memory updater, which is designed to selectively store and update the history dialog information so as to support downstream tasks. For the experiment, we employ human annotators to write and examine a large-scale dialog reading comprehension dataset. Extensive experiments are conducted on this dataset, and the results show that the proposed model brings substantial improvements over several strong baselines on both tasks. In this way, we demonstrate that reasoning can indeed help better response generation and vice versa. We release our large-scale dataset for further research.
The intelligent reflective surface (IRS) technology has received many interests in recent years, thanks to its potential uses in future wireless communications, in which one of the promising use cases is to widen coverage, especially in the line-of-s
ight-blocked scenarios. Therefore, it is critical to analyze the corresponding coverage probability of IRS-aided communication systems. To our best knowledge, however, previous works focusing on this issue are very limited. In this paper, we analyze the coverage probability under the Rayleigh fading channel, taking the number and size of the array elements into consideration. We first derive the exact closed-form of coverage probability for the unit element. Afterward, with the method of moment matching, the approximation of the coverage probability can be formulated as the ratio of upper incomplete Gamma function and Gamma function, allowing an arbitrary number of elements. Finally, we comprehensively evaluate the impacts of essential factors on the coverage probability, such as the coefficient of fading channel, the number and size of the element, and the angle of incidence. Overall, the paper provides a succinct and general expression of coverage probability, which can be helpful in the performance evaluation and practical implementation of the IRS.
To model diverse responses for a given post, one promising way is to introduce a latent variable into Seq2Seq models. The latent variable is supposed to capture the discourse-level information and encourage the informativeness of target responses. Ho
wever, such discourse-level information is often too coarse for the decoder to be utilized. To tackle it, our idea is to transform the coarse-grained discourse-level information into fine-grained word-level information. Specifically, we firstly measure the semantic concentration of corresponding target response on the post words by introducing a fine-grained focus signal. Then, we propose a focus-constrained attention mechanism to take full advantage of focus in well aligning the input to the target response. The experimental results demonstrate that by exploiting the fine-grained signal, our model can generate more diverse and informative responses compared with several state-of-the-art models.
In this paper, we investigate the impact of channel aging on the performance of cell-free (CF) massive multiple-input multiple-output (MIMO) systems with pilot contamination. To take into account the channel aging effect due to user mobility, we firs
t compute a channel estimate. We use it to derive novel closed-form expressions for the uplink spectral efficiency (SE) of CF massive MIMO systems with large-scale fading decoding and matched-filter receiver cooperation. The performance of a small-cell system is derived for comparison. It is found that CF massive MIMO systems achieve higher 95%-likely uplink SE in both low- and high-mobility conditions, and CF massive MIMO is more robust to channel aging. Fractional power control (FPC) is considered to compensate to limit the inter-user interference. The results show that, compared with full power transmission, the benefits of FPC are gradually weakened as the channel aging grows stronger.
Generating qualitative responses has always been a challenge for human-computer dialogue systems. Existing dialogue systems generally derive from either retrieval-based or generative-based approaches, both of which have their own pros and cons. Despi
te the natural idea of an ensemble model of the two, existing ensemble methods only focused on leveraging one approach to enhance another, we argue however that they can be further mutually enhanced with a proper training strategy. In this paper, we propose ensembleGAN, an adversarial learning framework for enhancing a retrieval-generation ensemble model in open-domain conversation scenario. It consists of a language-model-like generator, a ranker generator, and one ranker discriminator. Aiming at generating responses that approximate the ground-truth and receive high ranking scores from the discriminator, the two generators learn to generate improved highly relevant responses and competitive unobserved candidates respectively, while the discriminative ranker is trained to identify true responses from adversarial ones, thus featuring the merits of both generator counterparts. The experimental results on a large short-text conversation data demonstrate the effectiveness of the ensembleGAN by the amelioration on both human and automatic evaluation metrics.
With the development of self-interference (SI) cancelation technology, full-duplex (FD) communication becomes possible. FD communication can theoretically double the spectral efficiency. When the time slot (TS) resources are limited and the number of
flows is large, the scheduling mechanism of the flows becomes more important. Therefore, the effectiveness of FD scheduling mechanism for the flows is studied in millimeter wave wireless backhaul network with the limited TS resources. We proposed a full duplex concurrent scheduling algorithm based on coalition game (FDCG) to maximize the number of flows with their QoS requirements satisfied. We transformed the problem of maximizing the number of flows with their QoS requirements satisfied into the problem of maximizing sum rate of concurrently scheduled flows in each slot. We obtained the scheduled flows with maximum sum rate in first slot by using coalition game.And then with certain restrictions, the maximum sum rate of concurrently scheduled flows can also be achieved in subsequent time slots. The simulation results show that the proposed FDCG algorithm can achieve superior performance in terms of the number of flows that meet their QoS requirements and system throughput compared with other three algorithms.
