The selection of the spreading factor (SF) has important implications on the radio on-time, energy consumption, achievable data rate, and communication range of LoRa devices. In practical applications, LoRa packets can only be exchanged when the SF between transmitter and receiver matches. To ensure that this is the case, current approaches either statically hard-code the SF used to communicate between two devices, or negotiate which SF to use through handshaking mechanisms. Unfortunately, statically assigning the SF may lead to sub-optimal performance, and changing the assignment at runtime through a negotiation process incurs a significant overhead in terms of both latency and energy consumption. In this paper, we propose RSSF, a scheme that allows an off-the-shelf LoRa device to receive and decode a packet without prior knowledge of the SF used by the transmitter. RSSF leverages the observation that the SF with which a packet was sent can be inferred by analyzing the received signal strength (RSS) samples captured while receiving the first preamble symbols, and by characterizing their periodicity. In real-world systems, however, the waveform obtained by sampling the RSS during the reception of the first preamble symbols contains several spikes due to the receiver's DC offset cancellation, which makes it difficult to accurately identify periods. We show that this problem can be solved by letting an off-the-shelf LoRa receiver sample the RSS on a frequency that is shifted by half of the bandwidth from the original frequency at which the packet was transmitted. We then implement a lightweight algorithm that determines the SF by partitioning the RSS samples into sliding windows of different size (corresponding to each possible SF) and by measuring the zero-crossing intervals for each window size. We evaluate RSSF's performance experimentally using both software-defined radios and off-the-shelf LoRa nodes, showing that RSSF can accurately determine the SF within the first 5 preamble symbols.

LoRaWAN is one of the most popular protocols to build low-power wide area networks. Unfortunately, it adopts a star topology, which limits network coverage and may also cause an unnecessary decrease in energy efficiency as well as scalability. In fact, end-devices that are deployed far away from a gateway need to increase their transmission power or spreading factor (SF) to sustain reliable communications, which increases their energy expenditure as well as the size of the collision domain. The only alternative is the deployment of additional gateways or dedicated relay nodes, which results in higher costs and deployment efforts. In this work, we introduce LoRaHop, an extension of LoRaWAN that enriches end-devices with the ability to form a mesh network and to seamlessly relay packets to/from a gateway, thereby providing LoRaWAN networks with multihop support for both uplink and downlink messaging. LoRaHop leverages concurrent transmissions to enable a reliable and efficient data collection or dissemination over the mesh network, as well as to simplify network formation. Furthermore, LoRaHop embeds a mechanism that simplifies rendezvous across devices and that minimizes the impact of mesh operations on existing LoRaWAN transmissions. We implement LoRaHop on off-the-shelf LoRa end-devices (ensuring their interoperability with commercial LoRaWAN gateways and network servers), and evaluate its performance on an outdoor testbed. Our results show that LoRaHop can effectively extend the coverage of an LoRaWAN network while improving reliability by up to 98.33% and reducing energy consumption by up to 48.02%. Our findings further demonstrate that using LoRaHop to create a multihop LoRaWAN network that communicates using low SFs brings significant benefits in terms of energy efficiency and scalability compared to the use of a single-hop LoRaWAN network using high SFs.

This paper presents EMU, a framework that enables the emulation, snipping, and multiplexing of LoRa chirps on commercial IoT devices equipped with low-power sub-GHz transceivers, including those supporting LoRa itself. Chirp snipping consists in artificially removing a sequence of chips and in putting the radio in low-power mode, which allows to reduce energy consumption while still commu-nicating reliably. Chirp multiplexing exploits the gaps introduced by chirp snipping to transmit portions of another chirp on a sep-arate channel, which allows to concurrently transmit two LoRa packets and to increase the throughput. We build EMU as a modu-lar framework and implement support for off-the-shelf LoRa and non-LoRa transceivers. We then evaluate its performance by com-paring the reliability, efficiency, and receiver sensitivity achieved by EMU with that of traditional LoRa for different physical layer settings. We finally showcase EMU's ability to send packets over two channels simultaneously, thereby improving the uplink throughput of LoRaWan, and demonstrate that even non-LoRa transceivers employing EMU can communicate to a LoRaWan gateway, enabling new use cases and expanding the applicability of LoRa technology.

Mobile Edge Computing (MEC) can migrate traditionally deployed high-energy-consumption and high-complex computing tasks to nearby edge servers, providing users with faster services and better network performance. There is a security problem of malicious access by eavesdroppers when offloading computing tasks. In this article, we propose an access point (AP) and wireless devices (WDs) MEC system. WDs use a binary offloading strategy to calculate tasks. Taking the average time security and the stability of equipment buffer queue as constraints, an online computing offloading method based on Lyapunov is proposed. This method transforms the computing process into an online computing offloading problem with the goal of minimizing the total energy consumption of WDs, so as to decouple the CPU cycle frequency, transmission power and task offloading strategy in different time periods. Simulation results show that this method can achieve better performance in terms of energy consumption compared with other benchmark methods.

The Non-local self-attention mechanism can significantly improve the capability of feature representation with long-range dependencies at the cost of high computational complexity. To address the issue, the self-attention-based autoregressive axial transformer has been proposed to apply attention along a single axis of the feature maps instead of the whole ones with large receptive fields. It performs axial-attention twice along the height- and width-axis respectively in the spatial dimension of the feature maps for the image data. However, there is still room for improvement. We can convert the 2D spatial feature map into a 1D feature sequence and just perform axial-attention once along it to save more computing resources. Motivated by the insight, we propose an Efficient Axial-Attention Network (EAAN) for video-based person re-identification (Re-ID) to reduce computation and improve accuracy by serializing feature maps with multi-granularity and reducing the number of axial-attention runs. We also introduce a deserialization approach that can restore the shape of the feature maps. Moreover, we expand the CTN (Channel Transformer Network) to a wider range of uses. Additionally, we verify that the serialized feature sequence can enhance expressiveness in our EAAN with lower complexity. Experiments on MARS and DukeMTMC-VideoReID (DukeV) datasets show outstanding performance in computation efficiency and accuracy. It not only outperforms the state-of-the-art method on MARS by 0.3% in both Rank-1 and mAP, and surpasses that on DukeV by 0.1% in Rank-1 with equal mAP, but also reduces parameters and GFLOPS (Giga Floating-point Operations Per Second) by 16.9% and 6.6% respectively compared to another axial-attention-based method. The code will be available at https://github.com/hopstone/EAAN.

Recently, several datasets shedding light on connectivity aspects in real-world LoRa networks have been provided to the community. However, they typically only involve a limited number of nodes, deal with unidirectional communication only, or focus on very specific physical layer settings. More importantly, existing datasets typically lack fine-grained environmental information such as the temperature in the surroundings of each node, which is known to have a strong impact on communication performance. In this work, we provide the community with a comprehensive dataset that fills all these gaps. We have collected detailed connectivity information in an outdoor LoRa network composed of 21 nodes for more than four months. Our dataset does not only focus on network-level performance (e.g., the average number of correctly-exchanged packets), but sheds light on link-level information such as the received signal strength, signal-to-noise ratio, and the number of available neighbours over time. We further collect environmental information from an online weather site, as well as the on-board temperature of each node in the network, which varies considerably across the deployed locations. We collect all this information while perpetually changing physical layer settings such as the spreading factor and the RF channel. A preliminary analysis of our dataset, which is available in Zenodo1, reveals that temperature has a significant correlation with the link quality and connectivity in the outdoor LoRa network, confirming the findings of earlier studies.

The increasing congestion of the RF spectrum is a key challenge for low-power wireless networks using concurrent transmissions. The presence of radio interference can indeed undermine their dependability, as they rely on a tight synchronization and incur a significant overhead to overcome packet loss. In this paper, we present Harmony, a new data collection protocol that exploits the benefits of concurrent transmissions and embeds techniques to ensure a reliable and timely packet delivery despite highly congested channels. Such techniques include, among others, a data freezing mechanism that allows to successfully deliver data in a partitioned network as well as the use of network coding to shorten the length of packets and increase the robustness to unreliable links. Harmony also introduces a distributed interference detection scheme that allows each node to activate various interference mitigation techniques only when strictly necessary, avoiding unnecessary energy expenditures while finding a good balance between reliability and timeliness. An experimental evaluation on real-world testbeds shows that Harmony outperforms state-of-the-art protocols in the presence of harsh Wi-Fi interference, with up to 50% higher delivery rates and significantly shorter end-to-end latencies, even when transmitting large packets.

Concurrent transmission, a novel communication paradigm, has been shown to effectively achieve reliable and energy-efficient flooding in low-power wireless networks. With multiple nodes exploiting a receive-and-forward scheme, this technique works effectively in flooding-based networks, i.e., in one-to-many scenarios. However, application-level scheduling has to be introduced for data collection in wireless sensor networks (WSNs). In this paper, we propose Packet-in-Packet (PiP), an energy-efficient paradigm requiring no application-level scheduling for timely data collection in low-power WSNs. PiP builds on concurrent transmissions, exploiting constructive interference and the capture effect to achieve high reliability and low latency. Moreover, PiP uses a packet concatenation capability to gather single-hop information in a best-effort manner. As a result, PiP significantly reduces collection time. We further compare PiP with a state-of-the-art protocol by extensive experiments in a 30-node FlockLab testbed. Experimental results show that PiP highly reduces collection time (i.e., rounds) and achieves good performance in terms of high reliability of approximately 98.7% and high energy efficiency in all experimental scenarios of the real-world testbed.

Concurrent transmission, a novel communication paradigm, has been shown to effectively achieve reliable and energy-efficient flooding in low-power wireless networks. With multiple nodes exploiting a receive-and-forward scheme, this technique works effectively in one-to-many scenarios. However, application-level scheduling has to be introduced for data collection applications. In this paper, we propose Packet-in-Packet (PiP), an energy-efficient paradigm requiring no application-level scheduling for timely data collection in low-power wireless sensor networks. PiP builds on concurrent transmissions, exploiting constructive interference and the capture effect to achieve high reliability, low latency, and high energy efficiency. Moreover, PiP uses a packet concatenation capability to gather single-hop information in a best-effort manner. Experimental results show that PiP achieves high reliability in all experimental scenarios of a real-world testbed.

Vertical height estimation is critical to indoor localization technique. However, the common story height covers from 2.8m to 6.0m in multistory buildings, which make it meaningless to estimate height alone. An efficient indoor location system should provide accurate floor estimation with fuzzy story height information. This paper proposes a Bayesian Network inference method to identify pedestrian's floor level accurately in a multistory building with a waist-mounted device. The algorithm adopts an effective activities detector of stair climbing at first. With the output of the detector, the landing is counted and the height change is calculated by barometer measurements. Finally, based on the landing number and height change value, a Bayesian Network model is introduced to infer the floor change of the pedestrian. The experiments reveal that the proposed floor localization algorithm is more reliable, which achieves an accuracy of 99.36% with a total number of 1247 times floor change.

In this paper, we propose an algorithm for human activity recognition based on Gaussian Process Classifier (GPC). A hierarchical strategy is firstly applied to classify dynamic and static behaviors. Then, in each layer, three kinds of classification approaches are validated and evaluated for promot ing recognition accuracy. Moreover, discriminative analysis method is invoked to cast high dimension features into lower dimensional space where classes are easily separated. Extensive experiments have been conducted and three vital points are observed: Firstly, GPC achieves comparable classification accuracy with other classifiers under the same experimental condition. Secondly, in case of less training samples, GPC outperforms the prominent Support Vector Machine (SVM) classifier. Thirdly, unlike SVM, GPC is more robust to the high dimensional features. Furthermore, we successfully implement the presented recognition algorithm into our hardware platform and achieve 99.75% accuracy on average in dealing with four sample activities.

A new neural network (NN) aided adaptive unscented Kalman filter (UKF) is presented for tracking high maneuvering target. In practice, the dynamic systems of many target tracking problems are usually nonlinear and incompletely observed, moreover, there may be large modeling errors when the target is maneuverable or some parameters of the system models are inaccurate or incorrect. The adaptive capability of filters is known to be increased by incorporating a neural network into the filtering procedure. On the other hand, some nonlinear filtering methods such as extended Kalman Filter (EKF) have been used to train a NN with fast convergence speed by augmenting the state with unknown connecting weights. Tackling the natural coalescent between the filtering algorithm and the NN described above, first a more efficient learning algorithm based on unscented Kalman filter (UKF) is derived, which can give a more accurate estimate of the weights and possess faster convergence rate. We then extend the algorithm to form a new NN aided adaptive UKF algorithm and use it in maneuvering target tracking applications. The NN in this algorithm is used to approximate the uncertainty of system models and is trained online, together with the target state estimation. Some simulations are also given to validate that the proposed method can give well state estimation of a highly maneuvering target.

In this paper, a multi-model Rao-Blackwellised particle filter algorithm is presented for tracking high maneuvering target in distributed acoustic sensor networks. It is more efficient for high-dimension nonlinear and non-Gaussian estimation problems than generic particle filter, and by stratified particles sampling from a set of system models, it can tackle the target's maneuver perfectly. In the simulation comparison, a high maneuvering target moves through an acoustic sensor network field. The target is tracked using both the RBPF and the multi-model RBPF algorithms, and a location-central protocol is applied for energy conservation. The results show that our approach has great performance improvements, especially when the target is making maneuver.