Due to recent technological advancements in the area of unmanned aerial systems, equipping an unmanned aerial vehicle (UAV) with a base station (BS) has been proposed to augment terrestrial base stations and to enhance the performance of 5G and beyond-5G networks. In this paper, we investigate the 3D placement problem for multiple UAV-BSs that maximizes the number of covered users with the same as well as with different quality-of-service (QoS) requirements. First, we present a mathematical formulation of the multiple UAV-BS placement problem, and show that it is a non-convex optimization problem. Then, we propose two heuristic algorithms, and we show that for users with the same as well as with different QoS requirements, the proposed algorithms outperform the Linear Approximation (LA) state-of the-art algorithm in terms of the average number of covered users and execution time, and achieve near optimal performance. Finally, a tradeoff between the two heuristic algorithms in terms of the average number of covered users and their execution time is presented.

As the number of mobile devices that natively support ad hoc communication protocols increase, large ad hoc networks can be created not only to facilitate communication among the mobile devices, but also to assist devices that are executing computationally intensive applications. Prior work has developed computation offloading systems for mobile devices, but this work has focused exclusively on offloading to single hop neighbors, due in part to the practical challenges of setting up multi-hop networks using existing ad hoc communication protocols. However, limiting the offloading of computation to one-hop neighbors inherently restricts the number of devices that can participate in the distributed computation. By presenting a heuristic, aimed at avoiding partitioning the network, as well as an iterative task assignment algorithm that can optimize the assignment of computational tasks to devices in a multi-hop cooperative network, we are able to evaluate the effect of computational offloading in multi-hop networks. Experimental results, obtained from an implementation on Android devices, are integrated with an analytical model that enables the evaluation of system performance under a variety of conditions. These experimental and analytic results demonstrate the benefit of enabling computation offloading to all devices in a multi-hop cooperative network.

Wireless networks are vital for supporting a range of applications, and energy efficiency is a key factor in improving the operational lifetime for these networks. In multiantenna wireless networks, the energy optimization problem can be addressed using the trade-off between the transmit power and the circuit energy consumption. Multiple-Input Multiple-Output (MIMO) communication is a promising approach that can be efficiently used to reduce the energy consumption for communication. In MIMO systems, the transmit power is spread among more than one antenna, which results in having a high power gain and better spectral efficiency. In this paper, we propose optimizing the MIMO communication within a cluster-based protocol for wireless networks in which the nodes have multiple antennas and may be powered by energy harvesting. The proposed approach adjusts the number of antennas for communication between a normal node and a cluster-head dynamically in order to improve the overall energy efficiency of the network. We evaluate the proposed approach in two scenarios, a network with static nodes with non-rechargeable batteries, and a network with mobile nodes powered through energy harvesting. In both scenarios, the simulation results show that the proposed approach of multi-antenna optimization outperforms a traditional fixed approach in terms of number of received packets, percentage of dead nodes, and energy consumption.

The potential benefits of Mobile Ad Hoc Networks (MANETs) have led to the development of many protocols in order to control and optimize such networks. However, the automatic creation and evolution of ad hoc networks has yet to be exploited. A novel ad hoc protocol named WiFi Direct has been proposed and standardized by the WiFi Alliance with the objective of facilitating the interconnection of nearby devices. WiFi Direct provides high performance direct communication among devices and includes different energy management mechanisms. However, the current WiFi Direct implementations require user interaction for setting up and maintaining the connection. In this work, we exploit redundancy to enable the automatic and fast reconfiguration of WiFi Direct networks in response to network dynamics. We propose a proactive solution to unforeseen group owner failures in order to minimize the packet loss and network discontinuity time by setting up a redundant group on a second virtual network interface. Through emulation on Mininet-WiFi, we find that the proposed redundant scheme substantially decreases packet loss, providing almost continuous connectivity among nodes, which cannot be guaranteed through traditional WiFi Direct schemes. This is the first detailed work examining the auspicious potential of using an additional network interface to support network reformation using WiFi Direct.

The scarcity of emotional speech data is a bottleneck of developing automatic speech emotion recognition (ASER) systems. One way to alleviate this issue is to use unsupervised feature learning techniques to learn features from the widely available general speech and use these features to train emotion classifiers. These unsupervised methods, such as denoising autoencoder (DAE), variational autoencoder (VAE), adversarial autoencoder (AAE) and adversarial variational Bayes (AVB), can capture the intrinsic structure of the data distribution in the learned feature representation. In this work, we systematically investigate four kinds of unsupervised feature learning methods for improving speaker-independent ASER. We show that all methods improve the performance regarding unweighted accuracy rating (UAR) and Fl-score over methods that use hand-crafted features or that do not perform feature learning on external datasets. We also show that VAE, AAE and AVB methods, which control the distribution of the latent representation, outperform DAE that does not control such distribution. This suggests the benefits of using variational inference methods to learn features from general speech for the speech tasks such as ASER that has very limited labeled data.

Wireless Sensor Networks (WSNs) are crucial in supporting continuous environmental monitoring, where sensor nodes are deployed and must remain operational to collect and transfer data from the environment to a base-station. However, sensor nodes have limited energy in their primary power storage unit, and this energy may be quickly drained if the sensor node remains operational over long periods of time. Therefore, the idea of harvesting ambient energy from the immediate surroundings of the deployed sensors, to recharge the batteries and to directly power the sensor nodes, has recently been proposed. The deployment of energy harvesting in environmental field systems eliminates the dependency of sensor nodes on battery power, drastically reducing the maintenance costs required to replace batteries. In this article, we review the state-of-the-art in energy-harvesting WSNs for environmental monitoring applications, including Animal Tracking, Air Quality Monitoring, Water Quality Monitoring, and Disaster Monitoring to improve the ecosystem and human life. In addition to presenting the technologies for harvesting energy from ambient sources and the protocols that can take advantage of the harvested energy, we present challenges that must be addressed to further advance energy-harvesting-based WSNs, along with some future work directions to address these challenges.

Energy harvesting wireless nodes provide much longer lifetime and higher energy efficiency for wireless networks compared to battery operated systems. In this paper, we study a MIMO wireless communication link in which the nodes are equipped with energy harvesters and rechargeable batteries that are continuously charging from a renewable energy source. Since the harvested energy arrival and thus the future remaining energy of the nodes is not deterministic in practice, we propose a learning approach in order to find the most efficient transmission policy for data communication that maximizes throughput. The problem is formulated as a Markov Decision Process (MDP) with unknown transition probabilities. A Q-Learning approach is proposed to solve the MDP model and find the optimal transmission policy.

The wide diffusion of mobile devices that natively support ad hoc communication technologies has led to a number of protocols for enabling and optimizing Mobile Ad Hoc Networks (MANETs). Nevertheless, the actual utilization of MANETs in real life is still limited, in part due to the lack of protocols for the automatic creation and evolution of ad hoc networks. Recently, a novel ad hoc protocol named WiFi Direct has been proposed and standardized by the WiFi Alliance with the objective of facilitating the interconnection of nearby devices. WiFi Direct provides high performance direct communication among devices, includes different energy management mechanisms, and is now available in most modern mobile devices. However, the current WiFi Direct implementations require user interaction for setting up and maintaining the connection. In this paper, we propose and analyze three practical schemes for creating self-organizing and self-healing WiFi Direct networks using Android OS devices. Experimental results show that our proposed approaches are feasible with different overhead in terms of prior knowledge about the network and coordination between the devices. These techniques provide the first known approaches for the automatic creation and maintenance of MANETs using every day mobile devices.

Water is essential for human survival. Although approximately 71% of the world is covered in water, only 2.5% of this is fresh water; hence, fresh water is a valuable resource that must be carefully monitored and maintained. In developing countries, 80% of people are without access to potable water. Cholera is still reported in more than 50 countries. In Africa, 75% of the drinking water comes from underground sources, which makes water monitoring an issue of key concern, as water monitoring can be used to track water quality changes over time, identify existing or emerging problems, and design effective intervention programs to remedy water pollution. It is important to have detailed knowledge of potable water quality to enable proper treatment and also prevent contamination. In this article, we review methods for water quality monitoring (WQM) from traditional manual methods to more technologically advanced methods employing wireless sensor networks (WSNs) for in situ WQM. In particular, we highlight recent developments in the sensor devices, data acquisition procedures, communication and network architectures, and power management schemes to maintain a long-lived operational WQM system. Finally, we discuss open issues that need to be addressed to further advance automatic WQM using WSNs.

As the number of mobile devices that natively support ad hoc communication protocols increase, large ad hoc networks can be created not only to facilitate communication among the mobile devices, but also to assist devices that are executing computationally intensive applications. Prior work has developed computation offloading systems for mobile devices, but this work has focused exclusively on offloading to single hop neighbors, due in part to the practical challenges of setting up multi-hop networks using existing ad hoc communication protocols. However, limiting the offloading of computation to one-hop neighbors inherently restricts the number of devices that can participate in the distributed computation. In this paper, we propose and evaluate the performance of computational offloading within a multi-hop cooperative network, where mobile devices are able to share the computational load with all other nodes in the network. Additionally, we present an iterative task assignment algorithm that can optimize the assignment of computational tasks to devices in such a multi-hop cooperative network, taking into account the communication overhead of the multi-hop network. Experimental results, obtained from an implementation on Android devices, are integrated with an analytical model that enables the evaluation of system performance under a variety of conditions. These experimental and analytic results demonstrate the benefit of enabling computation offloading to all devices in a multi-hop cooperative network.

In areas where the habitats of elephants and humans are rapidly encroaching on each other, real-time monitoring of the elephants' locations has the potential to drastically improve the co-existence of elephants and humans, resulting in reduced deaths in both groups. However, as tagging (using GPS collars) elephants to obtain such location information is difficult and costly, it is important to ensure very long lifetimes of the tags, which can only be achieved using energy harvesting. In this paper, we present a kinetic energy harvester that uses magnetic levitation and ferro fluid bearings to generate energy from an elephant's movements. In order to determine the feasibility of using this kinetic energy harvester for powering the tags on elephants, we obtained real acceleration data collected from an Asian elephant over a 10 day period, and this data was then used to tune the system to maximize the harvested energy. Using experimentally validated analytical and simulation models, and the actual elephant acceleration data, we find that our prototype can generate 88.91J of energy per day. This energy is not only sufficient to power the tags to acquire and transmit locations 24 times a day to a distance of 114Km (line of sight), but provides a surplus of at least 35.40J, which can be used to increase the frequency of position updates or to support alternative communication options such as GPRS. Therefore, this shows the viability of long-term tracking of elephants.

The fact that emotions play a vital role in social interactions, along with the demand for novel human-computer interaction applications, have led to the development of a number of automatic emotion classification systems. However, it is still debatable whether the performance of such systems can compare with human coders. To address this issue, in this study, we present a comprehensive comparison in a speech-based emotion classification task between 138 Amazon Mechanical Turk workers (Turkers) and a state-of-the-art automatic computer system. The comparison includes classifying speech utterances into six emotions (happy, neutral, sad, anger, disgust and fear), into three arousal classes (active, passive, and neutral), and into three valence classes (positive, negative, and neutral). The results show that the computer system outperforms the naive Turkers in almost all cases. Furthermore, the computer system can increase the classification accuracy by rejecting to classify utterances for which it is not confident, while the Turkers do not show a significantly higher classification accuracy on their confident utterances versus unconfi-dent ones.

A passive wake-up radio in a wireless sensor network (WSN) has the advantage of increasing network lifetime by using a wake-up radio receiver (WuRx) to eliminate unnecessary idle listening. A sensor node equipped with a WuRx can operate in an ultra-low-power sleep mode, waiting for a trigger signal sent by the wake-up radio transmitter (WuTx). The passive WuRx is entirely powered by the energy harvested from radio transmissions sent by the WuTx. Therefore, it has the advantage of not consuming any energy locally, which would drain the sensor node's battery. Even so, the high amount of energy required to wake up a passive WuRx by a WuTx makes it difficult to build a multi-hop passive wake-up sensor network. In this paper, we describe and discuss our implementation of a battery-powered sensor node with multi-hop wake-up capability using passive WuRxs, called MH-REACH-Mote (Multi-hop-Range EnhAnCing energy Harvester-Mote). The MH-REACH-Mote is kept in an ultra-low-power sleep mode until it receives a wake-up trigger signal. Upon receipt, it wakes up and transmits a new trigger signal to power other passive WuRxs. We evaluate the wake-up range and power consumption of an MH-REACH-Mote through a series of field tests. Results show that the MH-REACH-Mote enables multi-hop wake-up capabilities for passive WuRxs with a wake-up range of 9.4m while requiring a reasonable power consumption for WuTx functionality. We also simulate WSN data collection scenarios with MH-REACH-Motes and compare the results with those of active wake-up sensor nodes as well as a low power listening approach. The results show that the MH-REACH-Mote enables a longer overall lifetime than the other two approaches when data is collected infrequently.

Power and energy consumption are the most important factors in extending the lifetime of Wireless Sensor Networks (WSN). Many energy efficiency techniques, that consider both the transmission and circuit power consumption have been proposed for the case of Single-Input Single-Output (SISO) WSNs. However, the power consumption of the receiver should also be considered in order to maximize the network lifetime. In this paper, we introduce a novel communication protocol for Multiple-Input Multiple-Output (MIMO) WSNs. In this protocol, the number of antennas to be used at both the transmitter and receiver are selected according to the energy consumption of the scheme, the remaining energy at the nodes, the distance between the nodes, and the target bit error rate. Starting from a policy that selects the optimal number of antennas, we then propose 3 low complexity heuristics with different information requirements. Numerical results show that our proposed communication protocols dramatically outperform the performance of a traditional fixed MIMO system in terms of energy consumption and system lifetime.

Mobile ad hoc networks (MANETs) are becoming increasingly common, and typical network loads considered for MANETs are increasing as applications evolve. This, in turn, increases the importance of bandwidth efficiency while maintaining tight requirements on energy consumption, delay and jitter. Coordinated channel access protocols have been shown to be well suited for highly loaded MANETs under uniform load distributions. However, these protocols are in general not as well suited for non-uniform load distributions as uncoordinated channel access protocols due to the lack of on-demand dynamic channel allocation mechanisms that exist in infrastructure based coordinated protocols. In this paper, we present a lightweight dynamic channel allocation mechanism and a cooperative load balancing strategy that are applicable to cluster based MANETs to address this problem. We present protocols that utilize these mechanisms to improve performance in terms of throughput, energy consumption and inter-packet delay variation (IPDV). Through extensive simulations we show that both dynamic channel allocation and cooperative load balancing improve the bandwidth efficiency under non-uniform load distributions compared to protocols that do not use these mechanisms as well as compared to the IEEE 802.15.4 protocol with GTS mechanism and the IEEE 802.11 uncoordinated protocol.

A wireless sensor network that employs passive radio wake-up of the sensor nodes can reduce the energy cost for unnecessary idle listening and communication overhead, extending the network lifetime. A passive wake-up radio is powered by the electromagnetic waves transmitted by a wake-up transmitter rather than a battery on the sensor node. However, this method of powering the wake-up radio results in a short wake-up range, which limits the performance of a passive wake-up radio sensor network. In this article, we describe our design of a passive wake-up radio sensor node—REACH 2 -Mote—using a high-efficiency, energy-harvesting module and a very low power wake-up circuit to achieve an extended wake-up range. We implemented REACH 2 -Mote in hardware and performed field tests to characterize its performance. The experimental results show that REACH 2 -Mote can achieve a wake-up range of 44 feet. We also modeled REACH 2 -Mote and evaluated its performance through simulations, comparing its performance to that of another passive wake-up radio approach, an active wake-up radio approach, and a conventional duty cycling approach. The simulation results show that REACH 2 -Mote can significantly extend the network lifetime while achieving high packet delivery rate and low latency.

As servers are placed in diverse locations in networked services today, it becomes vital to direct a client's request to the best server(s) to achieve both high performance and reliability. In this distributed setting, non-negligible latency and server availability become two major concerns, especially for highly-interactive applications. Profiling latencies and sending redundant data have been investigated as solutions to these issues. The notion of a cloudlet in mobile-cloud computing is also relevant in this context, as the cloudlet can supply these solution approaches on behalf of the mobile. In this paper, we investigate the effects of profiling and redundancy on latency when a client has a choice of multiple servers to connect to, using measurements from real experiments and simulations. We devise and test different server selection and data partitioning strategies in terms of profiling and redundancy. Our key findings are summarized as follows. First, intelligent server selection algorithms help find the optimal group of servers that minimize latency with profiling. Second, we can achieve good performance with relatively simple approaches using redundancy. Our analysis of profiling and redundancy provides insight to help designers determine how many servers and which servers to select to reduce latency.

Volunteer computing provides a practical and low cost solution to the increasing computational demands of many applications. Recent advancements in mobile device processing capabilities, combined with the energy efficiency of the mobile devices, make their inclusion in a distributed computing architecture particularly appealing. However, the intrinsic requirement of Internet connectivity to participate in volunteer computing limits the direct adoption of mobile devices due to service availability or related costs to connect to the Internet. In this paper, we propose and implement a novel computational architecture that extends the ability of mobile devices to participate in volunteer computing through ad hoc networking. By introducing decentralized task distribution points, mobile devices are invited to join the computation via device to device communication, removing the requirement for an Internet connection. Using a prototype implementation running on Android devices, we investigate the impact of a promising ad hoc communication technology, namely WiFi Direct, and two task distribution algorithms with different computation and communication overheads, under various scenarios. Experimental results show that our proposed approach is feasible with only minor additional energy consumption at the decentralized task distribution points.

Wireless transfer of energy through directed radio frequency waves has the potential to realize perennially operating sensor nodes by replenishing the energy contained in the limited on-board battery. However, the high power energy transfer from energy transmitters (ETs) interferes with data communication, limiting the coexistence of these functions. This paper provides the first experimental study to quantify the rate of charging, packet loss due to interference, and suitable ranges for charging and data communication of the ETs. It also explores how the placement and relative distances of multiple ETs affect the charging process, demonstrating constructive and destructive energy aggregation at the sensor nodes. Finally, we investigate the impact of the separation in frequency between data and energy transmissions, as well as among multiple concurrent energy transmissions. Our results aim at providing insights on radio frequency-based energy harvesting wireless sensor networks for enhanced protocol design and network planning.

Harvesting energy from radio frequency (RF) waves brings us closer to achieving the goal for perpetual operation of a wireless sensor network (WSN) by replenishing the batteries of the sensor nodes. However, due to restrictions on the maximum transmitted power, path loss, and receiver sensitivity, only a small amount of energy can be harvested. While a dedicated RF source alleviates the problem to some extent, novel techniques are required to boost the energy transfer efficiency of the source. In this paper, we provide the first experimental demonstration of multi-path energy routing (MPER) for the case of a sparsely distributed WSNs and show its improved performance over direct energy transfer (DET). In addition, we extend this concept to the case of densely distributed WSNs and experimentally demonstrate and compare the gains obtained by 2- and 3-path energy routing over DET. Our experimental results show that significant energy gains can be achieved in a dense network deployment even when the node to be charged is partially blocked by the neighboring nodes.

Fundamental frequency (F0) is one of the essential features in many acoustic related applications. Although numerous F0detection algorithms have been developed, the detection accuracy in noisy environments still needs improvement. We present a hybrid noise resilient F0detection algorithm named BaNa that combines the approaches of harmonic ratios and Cepstrum analysis. A Viterbi algorithm with a cost function is used to identify the F0value among several F0candidates. Speech and music databases with eight different types of additive noise are used to evaluate the performance of the BaNa algorithm and several classic and state-of-the-art F0detection algorithms. Results show that for almost all types of noise and signal-to-noise ratio (SNR) values investigated, BaNa achieves the lowest Gross Pitch Error (GPE) rate among all the algorithms. Moreover, for the 0 dB SNR scenarios, the BaNa algorithm is shown to achieve 20% to 35% GPE rate for speech and 12% to 39% GPE rate for music. We also describe implementation issues that must be addressed to run the BaNa algorithm as a real-time application on a smartphone platform.

Passive RFID systems used for inventory management and asset tracking typically utilize contention-based MAC protocols, such as the standard C1G2 protocol. Although the C1G2 protocol has the advantage that it is easy to implement, it suffers from unfairness and relatively low throughput when the number of tags in the network increases. This paper proposes a token-based MAC protocol called Token-MAC for passive RFID systems, which aims a) to provide a fair chance for tags in the network to access the medium without requiring synchronization of the tags, b) to increase the overall throughput, i.e., the tag rate, and c) to enable a high number of tags to be read under limited tag read time availability, which is an especially important challenge for mobile applications. We implement Token-MAC as well as C1G2 and a TDMA-based protocol using Intel WISP passive RFID tags and perform experiments. Additionally, based on our experimental results, we develop energy harvesting and communication models for tags that we then use in simulations of the three protocols. Our experimental and simulation results all show that Token-MAC can achieve a higher tag rate and better fairness than C1G2, and it can provide better performance over a longer range compared with the TDMA-based protocol. It is also shown that Token-MAC achieves much lower tag detection delay, especially for high numbers of tags. Token-MAC is, therefore, a promising solution for passive RFID systems.

Use of a passive wake-up radio can drastically increase the network lifetime in a sensor network by reducing or even completely eliminating unnecessary idle listening. A sensor node with a wake-up radio receiver (WuRx) can operate in an extremely low power sleep mode until it receives a trigger signal sent by a wake-up radio transmitter (WuTx). After receiving the trigger signal, the attached WuRx wakes up the sensor node to start the data communication. In this paper, we implement and compare the performance of three passive wake-up radio-based sensor nodes: 1) WISP-Mote, which is a sensor mote that employs an Intel WISP passive RFID tag as the WuRx; 2) EH-WISP-Mote, which combines a novel energy harvester with the WISP-Mote; and 3) REACH-Mote, which uses the energy harvester circuit combined with an ultra-low-power pulse generator to trigger the wake-up of the mote. Experimental results show that the wake-up range and wake-up delay for the EH-WISP-Mote are improved compared with the WISP-Mote, while providing the ability to perform both broadcast-based and ID-based wake-ups. On the other hand, the REACH-Mote, which can only provide broadcast-based wake-up, can achieve a much longer wake-up range than any known passive wake-up radio to date, achieving feasible wake-up at a range of up to 37 ft.

Batteries of field nodes in a wireless sensor network pose an upper limit on the network lifetime. Energy harvesting and harvesting aware medium access control protocols have the potential to provide uninterrupted network operation, as they aim to replenish the lost energy so that energy neutral operation of the energy harvesting nodes can be achieved. To further improve the energy harvesting process, there is a need for novel schemes so that maximum energy is harvested in a minimum possible time. Multi-hop radio frequency (RF) energy transfer is one such solution that addresses these needs. With the optimal placement of energy relay nodes, multi-hop RF energy transfer can save energy of the source as well as time for the harvesting process. In this work we experimentally demonstrate multi-hop RF energy transfer, wherein two-hop energy transfer is shown to achieve significant energy and time savings with respect to the single-hop case. It is also shown that the gain obtained can be translated to energy transfer range extension.

Cloud computing provides an approach to accessing shared computing resources. However, a traditional cloud is composed of powerful but energy-hungry workstations. The growth of the population of mobile devices such as smart phones and tablets provides huge amount of idling computing power. In this paper, we describe the design and implementation of a mobile computing system prototype named GEMCloud that utilizes energy efficient mobile devices (e.g., smartphones and tablets) as computing resources. We evaluate the computing power and energy efficiency of the mobile devices through comprehensive experiments. The results show that a cloud computing system with enough mobile devices working cooperatively is able to save 55% to 98% of the energy consumption of conventional server-based clouds while providing comparable computing speed.

Various applications of wireless sensor networks require multihop data transmission over error-prone wireless links, hence multipath routing can be used to meet the application's reliability requirement. On the other hand, wireless sensors are usually battery powered, and thus it is very important to save energy in order to prolong the network lifetime. In this article, we propose (1) Sleeping Multipath Routing, which can trade off reliability and network lifetime by dynamically activating an optimal number of paths to support the application's reliability requirement and putting the rest of the sensors to sleep, and (2) an adaptive sensor sleeping solution, which includes Sleeping Multipath Routing, a duty-cycled MAC protocol, and cross-layer coordination to further prolong the network lifetime by putting sensors to sleep at both the routing and the MAC layers. Our proposed Sleeping Multipath Routing and adaptive sensor sleeping solution can be implemented on any multipath routing protocol that discovers multiple disjoint paths between two nodes, and any duty-cycled MAC protocol that has fixed cycle length. We show an example of implementing our adaptive sensor sleeping solution using Directed Diffusion and S-MAC. Simulation results show that our proposed adaptive sensor sleeping solution outperforms single-layer sensor sleeping with a longer network lifetime with required reliability support. Moreover, our proposed adaptive sensor sleeping solution can significantly prolong the network lifetime when the application's reliability requirement varies over time, or when the number of alive nodes in the network changes over time.

We study the problem of joint optimization of Slepian-Wolf (SW) source coding and transmission rates over a Gaussian multiple access channel with the considerations of circuit power consumption and average transmit power constraint. The goal is to maximize the sample rate at the source nodes. We first derive a criterion to determine the optimality of different multiple access schemes such that the highest sample rate can be achieved at the source nodes when SW coding is used. Based on the derived optimality criterion, we propose a rate allocation procedure to determine the jointly optimal SW coding and transmission rates corresponding to orthogonal code division multiple access (CDMA), frequency division multiple access (FDMA), time division multiple access (TDMA) and superposition coding with successive interference cancellation (SCSIC) schemes. Several demonstrative numerical examples are provided to show the performance gain of the proposed joint rate allocation scheme.

An important aspect of the MAC layer design for mobile ad hoc networks (MANETs) is the ability to adapt to spatially non-uniform traffic load distributions. While coordinated channel access protocols have been shown to be well suited for highly loaded MANETs, these protocols are in general not as well suited for non-uniform load distributions as uncoordinated channel access protocols. In order to improve the performance of coordinated MAC protocols for MANETs under non-uniform load distribution, in this paper we apply to MANETs the dynamic channel allocation concept used in cellular networks. We present a new coordinated MAC protocol called DCA-TRACE (Dynamic Channel Allocation for TRACE), which is a sequel of our earlier MH-TRACE protocol. In addition to the mechanisms that exist in MH-TRACE for throughput and energy efficiency maximization, DCA-TRACE implements a distributed dynamic channel allocation strategy based on spectrum sensing. We compare DCA-TRACE with the MH-TRACE and 802.11 protocols in terms of throughput, energy consumption and interpacket delay variation (IPDV). Simulation results show that DCA-TRACE provides significant improvement for both uniform and non-uniform traffic loads.

Face recognition applications for airport security and surveillance can benefit from the collaborative coupling of mobile and cloud computing as they become widely available today. This paper discusses our work with the design and implementation of face recognition applications using our mobile-cloudlet-cloud architecture named MOCHA and its initial performance results. The challenge lies with how to perform task partitioning from mobile devices to cloud and distribute compute load among cloud servers (cloudlet) to minimize the response time given diverse communication latencies and server compute powers. Our preliminary simulation results show that optimal task partitioning algorithms significantly affect response time with heterogeneous latencies and compute powers. Motivated by these results, we design, implement, and validate the basic functionalities of MOCHA as a proof-of-concept, and develop algorithms that minimize the overall response time for face recognition. Our experimental results demonstrate that high-powered cloudlets are technically feasible and indeed help reduce overall processing time when face recognition applications run on mobile devices using the cloud as the backend servers.

Wireless image transmission is critical in many applications, such as surveillance and environment monitoring. In order to make the best use of the limited energy of the battery-operated cameras, while satisfying the application-level image quality constraints, cross-layer design is critical. In this paper, we develop an image transmission model that allows the application layer (e.g., the user) to specify an image quality constraint, and optimizes the lower layer parameters of transmit power and packet length, to minimize the energy dissipation in image transmission over a given distance. The effectiveness of this approach is evaluated by applying the proposed energy optimization to a reference ZigBee system and a WiFi system, and also by comparing to an energy optimization study that does not consider any image quality constraint. Evaluations show that our scheme outperforms the default settings of the investigated commercial devices and saves a significant amount of energy at middle-to-large transmission distances.

Emotion classification is essential for understanding human interactions and hence is a vital component of behavioral studies. Although numerous algorithms have been developed, the emotion classification accuracy is still short of what is desired for the algorithms to be used in real systems. In this paper, we evaluate an approach where basic acoustic features are extracted from speech samples, and the One-Against-All (OAA) Support Vector Machine (SVM) learning algorithm is used. We use a novel hybrid kernel, where we choose the optimal kernel functions for the individual OAA classifiers. Outputs from the OAA classifiers are normalized and combined using a thresholding fusion mechanism to finally classify the emotion. Samples with low `relative confidence' are left as `unclassified' to further improve the classification accuracy. Results show that the decision-level recall of our approach for six-class emotion classification is 80.5%, outperforming a state-of-the-art approach that uses the same dataset.

Wake-up radios present a promising alternative to traditional duty-cycle approaches, by reducing overhearing and by eliminating idle listening completely. However, energy-efficient operation of the wake-up mechanism is required for the success of such systems. In particular, addressing is an important component of the wake-up procedure that can consume a significant portion of the energy required. Hence, in this paper we propose a novel energy-efficient scheme for addressing. The proposed method is called Time-Knocking (TicK), since it utilizes the time duration of the MicroController Unit (MCU) sleeping to encode an address. In this paper, it is shown through mathematical analysis that TicK is much more energy efficient compared to the two common addressing mechanisms: MCU-decoding and correlator-based addressing. For realistic performance evaluations, physical measurements are performed to extract the parameter values of each method or the parameter values from well-documented off-the-shelf devices are used. TicK enables significant energy savings, especially in environments where false addressing initiations are frequent due to interference or where multiple nodes are addressed frequently.

Multicast routing protocols typically rely on the a priori creation of a multicast tree (or mesh), which requires the individual nodes to maintain state information. In dynamic networks with bursty traffic, where long periods of silence are expected between the bursts of data, this multicast state maintenance adds a large amount of communication, processing, and memory overhead for no benefit to the application. Thus, we have developed a stateless receiver-based multicast (RBMulticast) protocol that simply uses a list of the multicast members' (e.g., sinks') addresses, embedded in packet headers, to enable receivers to decide the best way to forward the multicast traffic. This protocol, called Receiver-Based Multicast, exploits the knowledge of the geographic locations of the nodes to remove the need for costly state maintenance (e.g., tree/mesh/neighbor table maintenance), making it ideally suited for multicasting in dynamic networks. RBMulticast was implemented in the OPNET simulator and tested using a sensor network implementation. Both simulation and experimental results confirm that RBMulticast provides high success rates and low delay without the burden of state maintenance.

Wireless sensor networks are often battery-powered, and hence extending the network lifetime is one of the primary concerns in the ubiquitous deployment of wireless sensor networks. One approach to efficiently utilize the limited energy supplies of the sensors is to have the medium access control (MAC) protocol duty-cycle the sensors, periodically putting the sensors to sleep and waking them up to reduce idle listening, which is energy intensive. Among duty-cycled MAC protocols, some protocols are synchronized so that nodes wake up at the same time in each cycle, and other protocols are asynchronous, where nodes have arbitrary offsets to start their cycles. For protocol designers, it is important to understand which type of duty-cycled MAC protocol should be chosen (synchronized or asynchronous), as well as what values should be assigned to the protocol parameters under a given network scenario in order to achieve a desirable performance for throughput, delay, or energy consumption. However, previous work to analyze the performance of different duty-cycled MAC protocols is either protocol-specific, or limited to one aspect of the performance metric. In this paper, we propose a Markov queuing model to analyze the throughput, delay, and energy consumption of both synchronized and asynchronous duty-cycled MAC protocols with applications to S-MAC and X-MAC. Our contributions include: 1) proposing a Markov queuing model to describe the queuing behavior of both synchronous and asynchronous duty-cycled nodes, 2) modeling the queue dynamics and the stationary probability of packet transmissions for S-MAC, a synchronized duty-cycled MAC protocol, to analyze its performance, 3) modeling the queue dynamics and the stationary probability of packet transmissions for X-MAC, an asynchronous duty-cycled MAC protocol, to analyze its performance, 4) providing comprehensive performance estimation and comparison for different duty-cycled MAC protocols, and 5) providing flexibility to tradeoff different performance metrics by optimizing the protocol parameters. Our model results are validated by comparing with NS-2 and Matlab simulations.

We consider an energy constrained wireless sensor network, with arbitrary number of nodes, where source nodes utilize Slepian-Wolf (SW) coding before transmission to a joint decoder. We investigate optimal and near-optimal SW coding rates, transmit powers, and transmit durations that maximize the number of collected samples during the network lifetime, subject to channel capacity, SW rate region, and residual energy constraints. We find optimal (near-optimal) closed-form solutions in the absence (presence) of an energy constraint at the joint decoder. We take into account the energy consumption of SW encoding and decoding and communication circuitry. Numerical results demonstrate the effectiveness of the proposed optimization, especially when the joint decoder is not energy constrained.

Passive RFID systems typically utilize contention-based MAC protocols, such as the standard C1G2 protocol. While these protocols are easy to implement, they suffer from unfairness and low throughput. This paper proposes a token-based MAC protocol called Token-MAC for passive RFID systems. Token-MAC is designed to enable fair access to the medium for all tags without requiring a-priori knowledge of the tags or synchronization of the tags. We implement Token-MAC as well as a TDMA-based protocol and compare these approaches with the standard C1G2 protocol. Our experiments show that Token-MAC has a higher tag rate and better fairness than C1G2, and can achieve good performance over a longer range than the TDMA-based protocol. Token-MAC is therefore a viable solution for many passive RFID systems.

Motion sensors and cameras can be used together to build in-home video monitoring systems, where the cameras are only activated when the motion sensors detect motion. This saves both bandwidth (reduced video storage and/or transmission) as well as energy (if the cameras are battery-operated). However, motion sensors and cameras have different fields of view (FoV), and thus it is not clear that attaching motion sensors to cameras provides the most efficient system. Therefore, it is necessary to evaluate the performance of systems where the motion sensors are attached to the cameras (attached scenario) as well as where the motion sensors are detached from the cameras (detached scenario). We consider the motion sensor and camera placement problem under these two scenarios, with an additional hybrid scenario, and we formulate optimization problems to achieve the minimum energy consumption, longest network lifetime, or lowest cost. Simulation results show the tradeoffs between different objectives for all three scenarios.

In wireless sensor networks, multipath routing is used to alleviate reliability degradation due to multihop transmissions over error-prone wireless channels. However, multipath routing is not energy-efficient, as it requires all the nodes in the network to always be awake. To prolong the network lifetime while maintaining a certain reliability performance, we propose Sleeping Multipath Routing, which selects the minimum number of disjoint paths to achieve a given reliability requirement and puts the rest of the network to sleep. Simulation results show that Sleeping Multipath Routing can significantly extend the network lifetime, and it can trade off reliability for lifetime.

We investigate the problem of maximizing the sample rate at the transmitters in a wireless network using distributed source coding (DSC) and lossless transmission. With the consideration of circuit power consumption, average power constraints and DSC rate constraints, we study different orthogonal multiple access channels (MACs) and prove that the optimal orthogonal MAC is either TDMA or CDMA, depending on the joint entropy of the correlated sources, average power constraints, circuit power consumption, noise power and path loss.

It is increasingly common to see multiple wireless networks existing in the same environment, being accessed by devices of the individual networks as well as by "hybrid" devices capable of communicating on multiple networks. Typically these networks do not cooperate or share resources, resulting in sub-optimal network performances. The efficiency of such networks can be increased by exploiting hybrid nodes to enable network interactions that allow cross-network performance optimizations and cross-network information and service sharing. Such mutually beneficial interactions of the networks is referred to as symbiotic networking. In this paper, we investigate the potential performance gains of such interactions for the routing protocols and the effect of the network parameters on these potential gains. We use both simulation and mathematical frameworks that model the routing performances of both symbiotic and independent networking to quantify the routing benefits of symbiotic networking. Our results show that symbiotic networking can provide vital support to co-located networks, which is especially important in resource-constrained networks such as wireless sensor networks or mobile ad hoc networks.

In this paper, we present Multicasting through Time Reservation using Adaptive Control for Energy efficiency (MC--TRACE), an energy-efficient real-time data multicasting architecture for mobile ad hoc networks. MC-TRACE is a cross-layer design, where the medium access control layer functionality and the network layer functionality are performed by a single integrated layer. The basic design philosophy behind the multicast routing part of the architecture is to establish and maintain an active multicast tree surrounded by a passive mesh within a mobile ad hoc network. Thus, the MC-TRACE multicast backbone is a condensed passive mesh woven around a highly pruned tree. Although tree- and mesh-based multicasting techniques have been used separately in existing multicasting architectures, the novelty in this study is the integration and reengineering of the tree and mesh structures to make them highly energy efficient and robust for real-time data multicasting in mobile ad hoc networks. Energy efficiency is achieved by enabling the nodes to switch to sleep mode frequently and by eliminating most of the redundant data receptions. We evaluated the performance of MC-TRACE through ns - 2 simulations and compared it with ODMRP. Our results show that MC-TRACE provides superior energy efficiency while producing competitive QoS performance and bandwidth efficiency.

As wireless sensor networks utilize battery-operated nodes, energy efficiency is of paramount importance at all levels of system design. In order to save energy in the transfer of data from the sensor nodes to one or more sinks, the data may be routed through other nodes rather than transmitting it directly to the sink(s). In this article, we investigate the problem of energy-efficient transmission of data over a noisy channel, focusing on the setting of physical-layer parameters. We derive a metric called the energy per successfully received bit, which specifies the expected energy required to transmit a bit successfully over a particular distance given a channel noise model. By minimizing this metric, we can find, for different modulation schemes, the energy-optimal relay distance and the optimal transmit energy as a function of channel noise level and path loss exponent. These results enable network designers to select the hop distance, transmit power, and/or modulation scheme that maximize network lifetime.

Energy efficiency is one of the crucial design criteria for wireless sensor networks. Idle listening constitutes a major part of energy waste, and thus solutions such as duty cycling and the use of wake-up radios have been proposed to reduce idle listening and save energy. Compared to duty cycling, wake-up radios save more energy by reducing unnecessary wake-ups and collisions. In this paper, we investigate the feasibility and potential benefits of using passive RFID as a wake-up radio. We first introduce a physical implementation of sensor nodes with passive RFID wake-up radios and measure their energy cost and wake-up probability. Then, we compare the performance of our RFID wake-up sensor nodes with duty cycling in a Data MULE scenario through simulations with realistic application parameters. The results show that using a passive RFID wake-up radio offers significant energy efficiency benefits at the expense of delay and the additional low-cost RFID hardware, making RFID wake-up radios beneficial for many delay-tolerant sensor network applications.

Network-wide broadcasting and multicasting are two important routing schemes used in group communications. In network-wide broadcasting, generated packets at the source node are distributed to all nodes in the network, while multicasting delivers the packets only to a subset of the nodes by creating and maintaining a data dissemination structure. Due to the overhead in multicasting, in certain situations, it is more efficient to use network-wide broadcasting instead of multicasting, even when the data is destined to a subset of the nodes. In this paper we analyze a specific broadcasting protocol and a specific multicasting protocol for two performance metrics, namely spectrum efficiency and energy efficiency, and determine the conditions that make one of them preferable over the other.

Energy conservation is one of the primary goals of the majority of MAC protocols designed for wireless sensor networks. In Sensor-MAC (S-MAC), one of the first MAC protocols designed for sensor networks, nodes sleep and wake up periodically to save energy. MAC protocols such as Timeout-MAC (T-MAC) and Advertisement-MAC (ADV-MAC) improved upon S-MAC by introducing support for variable loads and providing further energy savings. In event-driven networks with sources generating bursty traffic or networks with period traffic, we can further reduce the energy consumption and increase node lifetime if we take into account the nature of the traffic. In this paper we propose Advertisement-based Time-division Multiple Access, or ATMA, a distributed TDMA-based MAC protocol for wireless sensor networks that utilizes the bursty or periodic nature of the traffic to prevent energy waste through advertisements and reservations for data slots. We provide detailed comparisons of the ATMA protocol with S-MAC, T-MAC and ADV-MAC through extensive simulations. The simulation results show that in bursty or periodic traffics, ATMA outperforms S-MAC, T-MAC and ADV-MAC in terms of energy consumption with reductions of up to 75%, while faring as well as T-MAC and ADV-MAC in terms of packet delivery ratio and latency.

In this paper, we investigate the maximization of the amount of gathered data in a clustered wireless sensor network (WSN). The amount of gathered data is maximized by (1) choosing the optimal transmit power, and (2) selecting the optimal cluster head. For problem (1), we find closed-form solutions for the optimal or near optimal transmit power of cluster members (CM). For problem (2), we propose a near optimal cluster head selection (CHS) algorithm. The communication burden and computational complexity of CHS only grow linearly with the size of the cluster. In the proposed algorithms, iterations have been avoided in order to significantly lower the complexity of the algorithms compared with traditional iteration-based numerical optimization algorithms, making these approaches suitable for use in energy-constrained wireless sensor networks. The optimization gain is shown to be significant.

MAC layer duty-cycling is widely used to reduce idling listening, which is energy-intensive in wireless sensor networks. Among duty-cycling MAC protocols, asynchronous protocols generally have higher energy-efficiency and lower packet latency than synchronized ones. In this paper, we propose a Markov model to analyze the throughput of X-MAC, an asynchronous duty-cycled MAC protocol for wireless sensor networks. Simulations show that our analytical model provides throughput values that closely match the simulation results under various network conditions. Using our model to estimate the throughput of X-MAC is computationally efficient, providing accurate results quickly compared with simulations, which take a long time to run and require many iterations due to the large standard deviations in throughput. More importantly, our model enables designers to obtain a better understanding of the effect of different numbers of nodes and data arrival rates on the performance of X-MAC.

Duty cycling is often used to reduce the energy consumption caused by idle listening in Wireless Sensor Networks (WSNs). Most studies on WSN protocols define a common duty cycle value throughout the network to achieve synchronization among the nodes. On the other hand, a few studies propose adaptation of the duty cycle according to uniform traffic conditions, which is beneficial assuming one-to-one traffic patterns that result in evenly distributed packet traffic. In this work, we consider the convergecast communication pattern commonly observed in WSNs. In convergecast communication, the packet traffic observed around the sink node is much higher than the traffic observed far from the sink, i.e., nodes with different distances to the sink node receive and must relay different amounts of traffic. Additionally, we utilize receiver-based protocols, which enable nodes to communicate with no synchronization or neighbor information, and hence do not require all nodes in the network to have the same duty cycle. In this paper, we model the expected energy consumption of nodes utilizing receiver-based protocols as a function of their duty cycle and their distance to the sink node. Using this analysis, we derive a closed-form formula for the duty cycle that minimizes the expected energy consumption at a given distance. Moreover, we propose an adaptation method for the derived distance-based duty cycle, based on local observed traffic. Performance evaluations of the two proposed duty cycle assignment methods show that they greatly improve the energy efficiency without sacrificing packet delivery ratio or delay significantly.

In this paper, we propose an iteration-free algorithm to find the optimal configuration, including transmit power and source coding rates, to maximize the lifetime of a cluster utilizing Slepian-Wolf source coding of data sent to a fusion center. Exact closed form solutions are derived when the fusion center is not energy constrained. When the fusion center is energy constrained, a near optimum solution is provided. Numerical results demonstrating the performance of the proposed algorithms are also provided.

The 25 papers in this special issue focus on simple wireless sensor networking solutions.

The eight papers in this special section span across theoretical and practical considerations of various aspects of distributed camera networks, including adaptive sensing, distributed processing, efficient communications, and versatile implementations.

Many recent advances in MAC protocols for wireless sensor networks have been proposed to reduce idle listening, an energy wasteful state of the radio. Low-Power-Listening (LPL) protocols transmit packets for U s (the "interlistening interval"), thereby, allowing nodes to sleep for long periods of time between channel probes. The interlistening interval as well as the particular type of LPL protocol should be well matched to the network conditions. In this paper, we propose network-aware adaptation of the specific succession of repeated packets over the U interval (the "MAC schedule"), which yields significant energy savings. Moreover, some LPL protocols interrupt communication between the sender and the receiver after the data packet has been successfully received. We propose a new and simple adaptation of the "transmit/receive schedule" to synchronize nodes on a slowly changing path so that energy consumption and delay are further reduced, at no cost of overhead in most cases. Our results show that using network-aware adaptation of the MAC schedule provides up to 30 percent increase in lifetime for different traffic scenarios. Additional adaptation of the transmit/receive schedule to automatically synchronize the nodes can reduce packet delivery delays by up to 50 percent, providing an additional decrease in energy consumption of 18 percent.

Energy efficiency is of the utmost importance in wireless sensor networks. The family of Low-Power-Listening MAC protocols was proposed to reduce one form of energy dissipation-idle listening, a radio state for which the energy consumption cannot be neglected. Low-Power-Listening MAC protocols are characterized by a duty cycle: a node probes the channel every t_i {\rm s} of sleep. A low duty cycle favors receiving nodes because they may sleep for longer periods of time, but at the same time, contention may increase locally, thereby reducing the number of packets that can be sent. We propose two new approaches to control the duty cycle so that the target rate of transmitted packets is reached, while the consumed energy is minimized. The first approach, called asymmetric additive duty cycle control (AADCC), employs a linear increase/linear decrease in the t_i value based on the number of successfully received packets. This approach is easy to implement, but it cannot provide an ideal solution. The second approach, called dynamic duty cycle control (DDCC) utilizes control theory to strike a near-optimal balance between energy consumption and packet delivery successes. We generalize both approaches to multihop networks. Results show that both approaches can appropriately adjust t_i to the current network conditions, although the dynamic controller (DDCC) yields results closer to the ideal solution. Thus, the network can use an energy saving low duty cycle, while delivering up to four times more packets in a timely manner when the offered load increases.

Impulse Radio Ultra WideBand (IR-UWB) communication has proven to be an important technique for supporting high-rate, short-range, and low-power communication. In this paper, using detailed models of typical IR-UWB transmitter and receiver structures, we model the energy consumption per information bit in a single link of an IR-UWB system, considering packet overhead, retransmissions, and a Nakagami-m fading channel. Using this model, we minimize the energy consumption per information bit by finding the optimum packet length and the optimum number of RAKE fingers at the receiver for different transmission distances, using Differential Phase-shift keying (DBPSK), Differential Pulse-position Modulation (DPPM) and On-off Keying (OOK), with coherent and non-coherent detection. Symbol repetition schemes with hard decision (HD) combining and soft decision (SD) combining are also compared in this paper. Our results show that at very short distances, it is optimum to use large packets, OOK with non-coherent detection, and HD combining, while at longer distances, it is optimum to use small packets, DBPSK with coherent detection, and SD combining. The optimum number of RAKE fingers are also found for given transmission schemes.

Impulse radio ultra-wideband (IR-UWB) communication has proven an important technique for supporting high-rate, short-range, low-power communication. These are necessary criteria for emerging sensor networks, which oftentimes have very short distance communication requirements, necessitate low power operation and may require high data rates (e.g., for supporting the transmission of images or video). In this paper, using detailed models of typical IR-UWB transmitter/receiver structures, we minimize the energy consumption per information bit in a single link of an IR-UWB system, considering packet retransmissions and overhead. This minimization is realized by finding the optimum packet length and the optimum number of RAKE fingers at the receiver for different transmission distances, using differential BPSK (DBPSK) and OOK with coherent and non-coherent detection. Our results show that at very short distances, it is optimum to use OOK with non-coherent detection and large packets, and at longer distances, it is optimum to use DBPSK with coherent detection and small packets.

In this paper, we propose a complete protocol stack framework called UPS (Unified Protocol Stack) as a solution for incorporating different protocols within WSNs through enabling the co-existence of multiple modules in same stack layer as well as providing unified access to cross-layer data.

In this abstract, we present Advertisement MAC (ADVMAC), an advertisement based MAC protocol for wireless sensor networks. Simulation results show that ADV-MAC provides a better way to conserve energy and at the same time to support variable traffic loads.

Several medium access control (MAC) protocols have been proposed for wireless sensor networks with the objective of minimizing energy consumption. For example, sensor-MAC (S-MAC) was proposed to reduce energy consumption by introducing a duty cycle. Under variable traffic load, the fixed duty cycle of S-MAC results in energy loss due to idle listening along with higher latency and lower throughput compared to an adaptive duty cycle. Timeout-MAC (T-MAC) introduced such an adaptive duty cycle to handle variable traffic loads. However, nodes that do not take part in data exchange waste energy because of continuous renewal of their timeout values. In this paper we propose advertisement MAC (ADV-MAC), a MAC protocol for wireless sensor networks that eliminates this energy waste by introducing the concept of advertising for data contention. ADV-MAC minimizes the energy lost in idle listening while maintaining an adaptive duty cycle to handle variable loads. Additionally, ADV-MAC enables energy efficient MAC-level multicasting. We provide detailed comparisons of the ADV-MAC protocol with S-MAC and T-MAC through extensive simulations. The simulation results show that ADV-MAC efficiently handles variable load situations and provides substantial gains over S-MAC and T-MAC in terms of energy (reduction of up to 45%) while faring as well as T-MAC in terms of throughput and latency.

Multicast routing protocols typically rely on the a-priori creation of a multicast tree (or mesh), which requires the individual nodes to maintain state information. In sensor networks where traffic is bursty, with long periods of silence between the bursts of data, this multicast state maintenance adds a large amount of overhead for no benefit to the application. Thus, we have developed a stateless receiver-based multicast protocol that simply uses a list of the multicast members (e.g., sinks), embedded in packet headers, to enable receivers to decide the best way to forward the multicast traffic. This protocol, called RBMulticast (Receiver-Based Multicast), exploits the knowledge of the geographic locations of the nodes to remove the need for costly state maintenance (e.g., tree/mesh/neighbor table maintenance), making it ideally suited for sensor network multicast applications. RBMulticast was implemented in TinyOS and tested using a sensor network implementation as well as TOSSIM simulations. Both simulation and experimental results confirm that RBMulticast provides high success rates without the burden of state maintenance.

Soft clustering of the nodes combined with time division multiple access (TDMA) channel access within a cluster has been shown to provide an energy-efficient solution for mobile ad-hoc networks (MANET). Such channel access schemes use a parameter that is critical in determining network performance: the number of frames per superframe, which determines the amount of spatial reuse possible, similar to the frequency reuse factor in cellular networks. When a smaller number of frames per superframe is used, each frame will consist of a larger number of slots, enabling the frame (i.e., cluster) to support more nodes, but also limiting the choices of frames for clusterheads to select, causing higher co-channel interference and collisions. Conversely, when a larger number of frames per superframe is used, the clusterheads will only be able to grant channel access to a limited number of nodes, which in turn increases the number of dropped packets (i.e., blocked channel access). The optimum value of the number of frames is the one that minimizes the combined effect of both collisions and dropped packets. By analytically determining the effects of dropped packets and collisions, we can find the optimal value for any given scenario. This paper develops a model to determine the optimal TDMA structure under various settings, showing the advantages that can be obtained by adapting protocol parameters as network conditions change.

In mobile ad hoc networks (MANETs), Quality of Service (QoS) of a multicast protocol is one of the most important performance metrics. Channel conditions and network topology frequently change, and in order to achieve a certain QoS, complex algorithms and protocols are needed. Often channel conditions are neglected during the design of a multicast protocol. However, vulnerability against channel errors can severely cripple the performance of a multicast protocol. Mesh networking inspired multicasting approaches introduce increased redundancy in the routing process to overcome the performance loss due to channel errors. Although utilizing multiple paths from senders to receivers results in higher reliability, under better channel conditions the additional redundancy may not be needed in terms of reliability, and increased redundancy causes increased overhead. Therefore, we propose a mesh networking inspired approach that adapts the amount of redundancy according to the current link conditions. We show that this approach can achieve good QoS levels for real-time traffic scenarios while simultaneously reducing unnecessary energy dissipation.

Many sensor network applications require consistent coverage of the region in which they are deployed over the course of the network lifetime. However, because sensor networks may be deployed randomly, node distribution and data redundancy in some regions of the network may be lower than in others. The sensors in the sparsest regions should be considered more critical to the sensor network application since their removal would likely result in unmonitored regions in the environment. For this reason, sensors in the more densely deployed regions should be considered more favorable as candidates to route the traffic of other nodes in the network. In this work, we propose several coverage-aware routing costs that allow traffic to be routed around the sparsely deployed regions so that the coverage of the environment can remain high for a long lifetime. We also propose an integrated route discovery and sensor selection protocol called DAPR that further lengthens network lifetime by jointly selecting routers and active sensors, again with the goal of minimizing the use of sensors in sparsely covered areas. Simulation results show the effectiveness of our approach in extending network lifetime nearly to the extent that can be reached using a centralized approach based on global network knowledge.

Low-power-listening MAC protocols were designed to reduce idle listening, a major source of energy consumption in energy starved wireless sensor networks. Low-power-listening is a MAC strategy that allows nodes to sleep for tis (the ldquointer-listeningrdquo time) when there is no activity concerning them. It follows that a node has to occupy the medium for at least tis to guarantee that its destination will probe the channel at some point during the transmission. Low-power-listening protocols have evolved with the introduction of new radios, and the most recent contributions propose to interrupt communication between the sender and the receiver after the data packet has been successfully received and acknowledged. This results in significant energy savings because a sending node does not need to send for full tiperiods. We propose a new and simple approach to synchronize nodes on a slowly changing routing tree so that energy consumption is further reduced at the sending node, and the delay is considerably less. Our method allows the nodes to use a lower duty cycle, at no cost of overhead in most cases. Simulation and implementation results show that energy consumption can be reduced by a significant factor (dependant on ti) and delay by at least 18%.

Energy efficiency is of the utmost importance in wireless sensor networks. The family of low-power-listening MAC protocols was proposed to reduce one form of energy dissipation-idle listening, a radio state for which the energy consumption cannot be neglected. Low-power-listening (also called channel probing) MAC protocols are characterized by a duty cycle: a node probes the channel every tis of sleep. A low duty cycle favors receiving nodes because they may sleep for longer periods of time, but at the same time, contention may increase locally, thereby reducing the number of packets that can be sent. We propose a new approach to dynamically control the duty cycle so that the target rate of transmitted packets is reached, while the consumed energy is minimized. Our approach utilizes control theory and adapts it to the control of tifor low-power-listening MAC protocols in wireless sensor networks. Results show that this approach can appropriately adjust tito the current network conditions.

Quality of service (QoS) of a network-wide broadcast (NWB) protocol is one of the most important performance metrics, especially in mobile ad hoc networks, where channel conditions and network topology change frequently. We propose a mesh networking inspired approach to overcome the performance degradation caused by lossy channels. We show that our adaptive approach, whereby the amount of redundancy is adjusted to the current link conditions, can achieve good performance while simultaneously reducing unnecessary energy dissipation.

Single-hop centralized wireless sensor networks are widely used for applications ranging from security and surveillance to medical monitoring. Often the goal of these networks is to provide satisfactory quality of service (QoS) to the application under different system states, but it is difficult to determine how the appropriate sensor sets should be selected over time, given the knowledge of all possible sensor sets to support the application QoS requirements. To address this problem, in this paper we propose a novel cost function called SUI (Sensor Usage Index) that is based on a sensor's relative ideal lifetime (SRIL) and can be used to select sensor sets so as to meet application QoS requirements for extended periods of time. Simulation results show that utilizing the proposed cost function for sensor set selection enables the network to meet application QoS requirements for longer than using other standard sensor selection schemes. In fact, our scheme approaches the optimal network lifetime, which can be found using global knowledge of the sensors and the system dynamics.

In multihop wireless sensor networks that are often characterized by many-to-one (convergecast) traffic patterns, problems related to energy imbalance among sensors often appear. Sensors closer to a data sink are usually required to forward a large amount of traffic for sensors farther from the data sink. Therefore, these sensors tend to die early, leaving areas of the network completely unmonitored and reducing the functional network lifetime. In our study, we explore possible sensor network deployment strategies that maximize sensor network lifetime by mitigating the problem of the hot spot around the data sink. Strategies such as variable-range transmission power control with optimal traffic distribution, mobile-data-sink deployment, multiple-data-sink deployment, nonuniform initial energy assignment, and intelligent sensor/relay deployment are investigated. We suggest a general model to analyze and evaluate these strategies. In this model, we not only discover how to maximize the network lifetime given certain network constraints but also consider the factor of extra costs involved in more complex deployment strategies. This paper presents a comprehensive analysis on the maximum achievable sensor network lifetime for different deployment strategies, and it also provides practical cost-efficient sensor network deployment guidelines.

Wireless networks of visual sensors have recently emerged as a new type of sensor-based intelligent system, with performance and complexity challenges that go beyond that of existing wireless sensor networks. The goal of the visual sensor network we examine is to provide a user with visual information from any arbitrary viewpoint within the monitored field. This can be accomplished by synthesizing image data from a selection of cameras whose fields of view overlap with the desired field of view. In this work, we compare two methods for the selection of the camera-nodes. The first method selects cameras that minimize the difference between the images provided by the selected cameras and the image that would be captured by a real camera from the desired viewpoint. The second method considers the energy limitations of the battery powered camera-nodes, as well as their importance in the 3D coverage preservation task. Simulations using both metrics for camera-node selection show a clear trade-off between the quality of the reconstructed image and the network's ability to provide full coverage of the monitored 3D space for a longer period of time.

We examine the trade-off between lifetime and distortion in image sensor networks deployed for gathering visual information over a monitored region. Users navigate over the monitored region by specifying a viewpoint that varies with time, and the network attempts to meet the user requirement by synthesizing the desired view using a selection of cameras. We compare two camera selection methods, the first maximizes PSNR for the user's view without considering the cameras' available energy whereas the second uses knowledge of available energy at the cameras to maximize network lifetime. Our simulation results demonstrate a clear trade-off between the two strategies, with selection based on energy alone performing up to 3 dB worse initially than the PSNR based selection yet providing significantly higher coverage lifetime. In addition, we observe that under a fixed total energy constraint, more cameras with lower energy per node are preferable over fewer cameras with higher energy per node. Our results suggest that a hybrid or adaptive camera selection algorithm may provide the optimal lifetime-distortion trade-off.

We present a new dense multi-view registration technique for wide-baseline video/images that integrates a parametric optical flow-based approach with a sparse set of feature correspondences, based on a locally planar approximation of a nonplanar scene. The proposed method can deal with illuminance variations between the views, which is critically important for wide-baseline applications. It differs from existing work on wide-baseline image registration in that it requires only image information and provides dense matching without computing any camera calibration matrices or performing any prior scene segmentation. These characteristics render the method suitable for practical deployment in visual sensor networks, towards which the current work is directed. We demonstrate the performance of the proposed method on simulated multi-view images of a virtual 3D world composed of piece-wise smooth textured surfaces, as well as real wide-baseline images of nonplanar textured surfaces.

We propose a new adaptive filtering framework for local image registration, which compensates for the effect of local distortions/displacements without explicitly estimating a distortion/displacement field. To this effect, we formulate local image registration as a two-dimensional (2-D) system identification problem with spatially varying system parameters. We utilize a 2-D adaptive filtering framework to identify the locally varying system parameters, where a new block adaptive filtering scheme is introduced. We discuss the conditions under which the adaptive filter coefficients conform to a local displacement vector at each pixel. Experimental results demonstrate that the proposed 2-D adaptive filtering framework is very successful in modeling and compensation of both local distortions, such as Stirmark attacks, and local motion, such as in the presence of a parallax field. In particular, we show that the proposed method can provide image registration to: a) enable reliable detection of watermarks following a Stirmark attack in nonblind detection scenarios, b) compensate for lens distortions, and c) align multiview images with nonparametric local motion.

In this paper, we present NB-TRACE, which is an energy-efficient network-wide voice broadcasting architecture for mobile ad hoc networks. In the NB-TRACE architecture, the network is organized into overlapping clusters through a distributed algorithm, where the clusterheads create a nonconnected dominating set. Channel access is regulated through a distributed TDMA scheme maintained by the clusterheads. The first group of packets of a broadcast session is broadcast through flooding, where each data rebroadcast is preceded by an acknowledgment to the upstream node. Nodes that do not get an acknowledgment for a predetermined time, except the clusterheads, cease to rebroadcast, which prunes the redundant retransmissions. The connected dominating set formed through this basic algorithm is broken in time due to node mobility. The network responds to the broken links through multiple mechanisms to ensure the maintenance of the connected dominating set. We compare NB-TRACE with four network layer broadcast routing algorithms (flooding, gossiping, counter-based broadcasting, and distance-based broadcasting) and three medium access control protocols (IEEE 802.11, SMAC, and MH-TRACE) through extensive ns-2 simulations. Our results show that NB-TRACE outperforms other network/MAC layer combinations in minimizing energy dissipation and optimizing spatial reuse, while producing competitive QoS performance

Image registration is a fundamental task in both image processing and computer vision. We present a novel method for local image registration based on adaptive filtering techniques. We utilize an adaptive filter to estimate and track correspondences among multiple images containing overlapping views of common scene regions. Image pixels are traversed in an order established by space-filling curves, to preserve the contiguity and hence track locally varying registration changes. The algorithm differs from pre-existing work on image registration in that it requires only local information and relatively low computational effort. These characteristics render the method suitable for deployment in imaging sensor networks, toward which the current work is directed. We evaluate the performance of the proposed algorithm using images captured with a digital camera in various real-world scenarios. Experimental results show that the proposed method can significantly improve accuracy and robustness over a global 2D parametric registration and can also outperform the local registration algorithm based on the Lucas-Kanade optical flow technique (Lucas, B. and Kanade, T., 1981).

Car crashes claim the lives of more than 100,000 people every year in the US alone. Forming ad-hoc networks among vehicles traveling on a highway can be very helpful to avoid such deadly accidents and pile-ups. In this paper, we define two classes of applications for such networks: safety-related applications and Internet connectivity. We also propose a new model for highway traffic and events that can be used to automatically generate movement files readable by the NS-2.28 simulator. Through simulations of such vehicular networks using flooding and IEEE 802.11 for safety-related applications, we attempt to answer the fundamental question: are highway vehicular networks feasible and efficient for safety purposes

In multi-hop wireless sensor networks that are characterized by many-to-one (converge-east) traffic patterns, problems related to energy imbalance among sensors often appear. When the transmission range is fixed for nodes throughout the network, the amount of traffic that sensors are required to forward increases dramatically as the distance to the data sink becomes smaller. Thus, sensors closest to the data sink tend to die early. Network lifetime can be improved to a limited extent by the use of a more intelligent transmission power control policy that balances the energy used in each node by requiring nodes further from the data sink to transmit over longer distances (although not directly to the data sink). Alternatively, policies such as data aggregation allow the network to operate in a more energy efficient manner. Since the deployment of an aggregator node may be significantly more expensive than the deployment of an ordinary microsensor node, there is a cost tradeoff involved in this approach. This paper provides an analysis of these policies for mitigating the sensor network hot spot problem, considering energy efficiency as well as cost efficiency.

We present network-wide broadcasting through time reservation using adaptive control for energy efficiency (NB-TRACE), which is an energy-efficient network-wide voice broadcasting architecture for mobile ad hoc networks. In the NB-TRACE architecture, the network is organized into overlapping clusters, where the clusterheads create a non-connected dominating set. Channel access is regulated through a locally maintained distributed TDMA scheme. The first group of packets of a broadcast session is broadcast through blind flooding. Each data rebroadcast includes an implicit acknowledgement to the upstream node. Nodes that do not get acknowledgement for a predetermined time, except the clusterheads, cease to rebroadcast, which prunes redundant retransmissions. The distributed connected dominating set formed through this basic algorithm is broken in time due to node mobility. The network responds to the broken links through passive and active clusterhead data transmission monitoring to ensure the maintenance of the connected dominating set. We compare NB-TRACE with flooding and gossiping using MH-TRACE, IEEE 802.11, and SMAC medium access control protocols through ns-2 simulations. Our results show that NB-TRACE outperforms other network/MAC layer combinations in terms of energy efficiency, packet delivery ratio, jitter, and number of rebroadcasts.

In this paper, we investigate the effects of channel noise on the performance of coordinated and non-coordinated MAC protocols. Comparative evaluations of these protocols under a perfect channel assumption have shown that coordinated MAC protocols, which regulate channel access locally, outperform non-coordinated channel access schemes in terms of energy efficiency and throughput. However, coordinated MAC protocols are more vulnerable than non-coordinated MAC protocols to channel noise due to their dependence on the robustness of the control traffic. In order to observe the degradation in performance of a coordinated MAC protocol (MH-TRACE), we investigate the impact of losing control packets. Furthermore, the performance in terms of throughput, delay, and energy efficiency of both coordinated (MH-TRACE) and non-coordinated (IEEE 802.11) MAC protocols is explored using a general error model that takes into account the length of the packets. Our results show that despite its higher level of vulnerability, the coordinated MAC protocol's performance is superior to the performance of the non-coordinated MAC protocol even when error rates are high.

Routing protocols for mobile ad hoc networks (MANETs) have been explored extensively in recent years. Much of this work is targeted at finding a feasible route from a source to a destination without considering current network traffic or application requirements. Therefore, the network may easily become overloaded with too much traffic and the application has no way to improve its performance under a given network traffic condition. While this may be acceptable for data transfer, many real-time applications require quality-of-service (QoS) support from the network. We believe that such QoS support can be achieved by either finding a route to satisfy the application requirements or offering network feedback to the application when the requirements cannot be met. We propose a QoS-aware routing protocol that incorporates an admission control scheme and a feedback scheme to meet the QoS requirements of real-time applications. The novel part of this QoS-aware routing protocol is the use of the approximate bandwidth estimation to react to network traffic. Our approach implements these schemes by using two bandwidth estimation methods to find the residual bandwidth available at each node to support new streams. We simulate our QoS-aware routing protocol for nodes running the IEEE 802.11 medium access control. Results of our experiments show that the packet delivery ratio increases greatly, and packet delay and energy dissipation decrease significantly, while the overall end-to-end throughput is not impacted, compared with routing protocols that do not provide QoS support.

We present a quality of service (QoS) architecture for supporting real-time data transmission in mobile ad hoc networks (MANETs). The QoS architecture includes a QoS transport layer, QoS routing, queue management and a priority MAC protocol. Through simulations, we find that the QoS architecture reduces packet delay and greatly improves the quality of real-time video streams in MANETs.

In this work, we present a method for the selection of a subset of nodes in a wireless sensor network whose application is to reconstruct the image of a (spatially) bandlimited physical value (e.g., temperature). The selection method creates a sampling pattern based on blue noise masking and guarantees a near minimal number of activated sensors for a given signal-to-noise ratio. The selection method is further enhanced to guarantee that the sensor nodes with the least residual energy are the primary candidates for deselection, while enabling a tradeoff between sensor selection optimality and balanced load distribution. Simulation results show the effectiveness of these selection methods in improving signal-to-noise ratio and reducing the necessary number of active sensors compared with simpler selection approaches.

In mobile ad hoc networks, node mobility causes links between nodes to break frequently, thus terminating the lifetime of the routes containing those links. An alternative route has to be discovered once a link is detected as broken, incurring extra route discovery overhead and packet latency. A simple solution to reduce the frequency of this costly discovery procedure is to choose a long lifetime route carefully during the route discovery phase rather than a simple random shortest-path route scheme. This simple solution, although straightforward, requires investigation of several questions before it can be implemented. Specifically, how much effect does node mobility have on the link lifetime? Correspondingly, how much effect does it have on the route lifetime? How much lifetime extension can we achieve by using a long lifetime route compared to a randomly chosen shortest-path route? How much benefit can we obtain by using these longer lifetime routes, and what is the tradeoff of using them? By answering these questions, we gain enough insight to determine under what circumstances long lifetime routes are worth being discovered and implemented. In this paper, we first formulate the distribution of a link lifetime and correspondingly the distribution of a route lifetime. Then we present an algorithm to determine the longest lifetime routes at different route lengths. We experimentally compare these long lifetime routes with traditional random shortest-path routes and reveal the tradeoff of using long lifetime routes of different lengths. All the fundamental results will serve as the guideline for our future implementation of a distributed long lifetime routing protocol.

As wireless sensor networks continue to attract more attention, new ideas for applications are continually being developed, many of which involve consistent coverage of a given surveillance area. Recently, several protocols and architectures have been proposed to maintain network connectivity and adequate coverage quality while minimizing the drain on the scarce energy resources of the sensor nodes. In this paper, we propose DAPR, an integrated protocol for routing and coverage preservation that is distinctly different from the previously proposed solutions. Guided by the intuition that certain sensors are more important to the sensing application than others because of limited neighborhood redundancy, we introduce a new routing metric-an "application cost"-that aims to avoid the use of sensors in areas of critically sparse sensor deployment as routers. We have implemented DAPR in the ns-2 simulator and present simulation results showing the effectiveness of the protocol in extending network lifetime.

In this paper, we propose multihop time reservation using adaptive control for energy efficiency (MH-TRACE), which is a medium access control (MAC) protocol that combines advantageous features of fully centralized and fully distributed networks for energy-efficient real-time packet broadcasting in a multihop radio network. We introduce a novel clustering algorithm that dynamically organizes the network into two-hop clusters. MH-TRACE clusters are just for coordinating channel access and minimizing interference; thus, ordinary nodes are not static members of any cluster. Time is organized into cyclic superframes, which consist of several time frames, to support reservation-based periodic channel access for real-time traffic. Each clusterhead chooses the frame with least interference based on its own measurements for the operation of its cluster. Energy dissipation for receiving unwanted or collided data packets or for waiting in idle mode is avoided through the use of information summarization packets sent prior to the data transmissions by the source nodes. Through the use of transmission schedules within each cluster, managed by the clusterheads, intracluster data collisions are completely eliminated and intercluster collisions are minimized. We investigated MH-TRACE through extensive simulations and theoretical analysis. Our results show that MH-TRACE outperforms existing distributed MAC protocols like IEEE 802.11 and sensor MAC, in terms of energy efficiency and throughput, approaching the theoretical maximum throughput and theoretical minimum energy dissipation.

In many surveillance video applications, it is of interest to recognize an object or a person, which occupies a small portion of a low-resolution, noisy video. This paper addresses the problem of super-resolution recovery of a region of interest from more than one low-resolution view of a scene recorded by multiple cameras. The multiple camera scenario alleviates the difficulty in registration of multiple frames of video that contain non-rigid or multiple object motion in the single camera case. With proper temporal registration of multiple videos, arbitrary scene motion can be handled. The success of super-resolution recovery from multiple views in real applications vitally depends on two factors: i) the accuracy of multiple view registration results, and ii) the accuracy of the camera and data acquisition model. We propose a system, which consists of a method for sub-pixel accurate spatio-temporal alignment of multiple video sequences for view registration and the projections onto convex sets method for super-resolution recovery. Experiments were implemented using two commercial analog video cameras, which do not perform on-board compression. Experimental results show that the super resolution recovery of dynamic scenes can be achieved as long as the multiple views of the scene can be registered with sub-pixel accuracy.

Future smart environments are characterized by multiple nodes that sense, collect, and disseminate information about environmental phenomena through a wireless network. In this paper, we define a set of applications that require a new form of distributed knowledge about the environment, referred to as non-uniform information granularity. By non-uniform information granularity we mean that the required accuracy or precision of information is proportional to the distance between a source node (information producer) and current sink node (information consumer). That is, as the distance between the source node and sink node increases, loss in information precision is acceptable. Applications that can benefit from this type of knowledge range from battlefield scenarios to rescue operations. The main objectives of this paper are two-fold: first, we precisely define non-uniform information granularity, and second we describe the different protocols that achieve non-uniform information dissemination and analyze these protocols based on complexity, energy consumption, and accuracy of information.

In this paper, we address a fundamental problem concerning the best flooding strategy to minimize cost and latency for target discovery in wireless networks. Should we flood the network only once to search for the target, or should we apply a so-called "expansion ring" mechanism to reduce the cost? If the "expansion ring" mechanism is better in terms of the average cost, how many rings should there be and what should be the radius of each ring? We separate wireless networks based on network scale and flooding control methods and explore these questions. We prove that when using a geography-based flooding control method, the number of flooding attempts should be one. When using a hop-based flooding control method, we prove that two-tier and three-tier schemes can reduce the cost of flooding compared to a single attempt. We provide a general formula to determine good parameters for the two-tier and threetier hop-based flooding schemes. Through simulations, we show that choosing flooding parameters according to our techniques gives performance close to that of ideal flooding schemes. Also, we present some preliminary results on flooding strategy with caching and show that a properly chosen searching radius can save much more query overhead than a simple radius selection scheme.

In this work, we address the problem of maximizing the lifetime of an application that requires a minimum level of quality of service from a network of energy-constrained wireless sensors. The problem is formulated as a generalized maximum flow graph problem with additional constraints and an optimal solution is found through linear programming. The result of the optimization is a schedule that determines the mode that all sensors should operate in and allows redundant sensors to turn off and conserve energy whenever possible. We show through simulations of typical sensing applications that by intelligently managing sensors, application lifetime can be extended by up to a factor of 2 in some cases.

Time reservation using adaptive control for energy efficiency (TRACE) is a time frame based media access control (MAC) protocol designed primarily for energy-efficient reliable real-time voice packet broadcasting in a peer-to-peer, single-hop infrastructureless radio network. Such networks have many application areas for various scenarios that obey a strongly connected group mobility model, such as interactive group trips, small military or security units, and mobile groups of hearing impaired people. TRACE is a centralized MAC protocol that separates contention and data transmission, providing high throughput, bounded delay, and stability under a wide range of data traffic. Furthermore, TRACE uses dynamic scheduling of data transmissions and data summarization prior to data transmission to achieve energy efficiency, which is crucial for battery operated lightweight radios. In addition, energy dissipation is evenly distributed among the nodes by switching network controllers when the energy from the current controller is lower than other nodes in the network, and reliability is achieved through automatic controller backup features. TRACE can support multiple levels of quality-of-service, and minimum bandwidth and maximum delay for voice packets are guaranteed to be within certain bounds. In this paper, we describe TRACE in detail and evaluate its performance through computer simulations and theoretical analysis.

We propose and analyze a routing protocol for mobile ad hoc networks that supports efficient image retrieval based on metadata queries. In digital photography, metadata describes captured information about an image and provides the key element needed for advanced techniques for sharing pictures. Our goal was to find an efficient way to utilize metadata to retrieve images in a wireless network of imaging devices. Building on the SPIN (sensor protocols for information via negotiation) protocol for metadata negotiation, we designed SPIN-IT (SPIN-image transfer), a protocol where wireless imaging devices use metadata queries to retrieve desired pictures. This protocol provides low bandwidth query-based communication prior to the transfer of image data to set up routes to desired data rather than routes to specific nodes. We compare SPIN-IT to a centralized approach and discuss the advantages of each design for different picture-sharing scenarios.

Networking together hundreds or thousands of cheap microsensor nodes allows users to accurately monitor a remote environment by intelligently combining the data from the individual nodes. These networks require robust wireless communication protocols that are energy efficient and provide low latency. We develop and analyze low-energy adaptive clustering hierarchy (LEACH), a protocol architecture for microsensor networks that combines the ideas of energy-efficient cluster-based routing and media access together with application-specific data aggregation to achieve good performance in terms of system lifetime, latency, and application-perceived quality. LEACH includes a new, distributed cluster formation technique that enables self-organization of large numbers of nodes, algorithms for adapting clusters and rotating cluster head positions to evenly distribute the energy load among all the nodes, and techniques to enable distributed signal processing to save communication resources. Our results show that LEACH can improve system lifetime by an order of magnitude compared with general-purpose multihop approaches.

Wireless microsensor networks lend themselves to trade-offs in energy and quality. In these networks, the individual sensor data per se are not necessarily important to the end user. Rather, it is the combined knowledge of all the sensors that describes what is occurring in the environment. By allowing the algorithms and protocols to adapt the quality of this description, with a corresponding change in energy dissipation, sensor networks can be flexible to the end-user's requirements. In this paper, we provide models for predicting quality and energy and show the advantages of trading off these two parameters. By ensuring that the system operates at a minimum energy for each quality point, the system can achieve both flexibility and energy efficiency, allowing the end-user to maximize system lifetime.

Technology has advanced in recent years to the point where multimedia communicators are beginning to emerge. These communicators are low-power, portable devices that can transmit and receive multimedia data through the wireless network. Due to the high computational complexity involved and the low-power constraint in wireless applications, these devices require the use of processors that are powerful and are at the same time very power-efficient. In order to facilitate interoperability, it is important that these devices use standardized compression and communication algorithms. As a first step in implementing multimedia terminals, Texas Instruments (TI) has demonstrated real-time MPEG-4 video decoding (simple profile) on TMS320C54x, TI's low power, high performance DSP chip. In addition, TI has outlined a system-level solution to transmitting video across wireless networks, including channel coding and communication protocols.

The MPEG-4 video compression standard incorporates several techniques to create error-resilient coded video. These techniques are effective against a certain level of errors. However, wireless channels often have very high error rates; thus channel coding is needed to reduce the number of errors in the compressed bitstream that is sent to the MPEG-4 video decoder. The structure of an MPEG-4 compressed bitstream lends itself to using unequal error protection to ensure fewer errors in the important portions of the bitstream. This paper discuses how unequal error protection can be used with the MPEG-4 error resilience tools and describes several experiments using both unequal and equal error protection on video sent through a simulated GSM channel. The results of these experiments show that when the bit error rate is high, unequal error protection can improve reconstructed video quality by as much as 1 dB compared with equal error protection.

In this paper, we present a family of adaptive protocols, called SPIN (Sensor Protocols for Information via Negotiation), that efficiently disseminates information among sensors in an energy-constrained wireless sensor network.Nodes running a SPIN communication protocol name their data using high-level data descriptors, called meta-data.They use meta-data negotiations to eliminate the transmission of redundant data throughout the network.In addition, SPIN nodes can base their communication decisions both upon application-specific knowledge of the data and upon knowledge of the resources that are available to them.This allows the sensors to efficiently distribute data given a limited energy supply.We simulate and analyze the performance of two specific SPIN protocols, comparing them to other possible approaches and a theoretically optimal protocol.We find that the SPIN protocols can deliver 60% more data for a given amount of energy than conventional approaches.We also find that, in terms of dissemination rate and energy usage, the SPlN protocols perform close to the theoretical optimum.

Motion estimation has been shown to help significantly in the compression of video sequences. However, since most motion estimation algorithms require a large amount of computation, it is undesirable to use them in power constrained applications, such as battery operated wireless video terminals. This paper presents an approach to reducing the power dissipation of wireless video terminals in a networked environment by exploiting the predictability of object motion. Since the location of an object in the current frame can be predicted from its location in previous frames, it is possible to optimally partition the motion estimation computation between battery operated portable devices and high powered compute servers on the wired network. This can achieve a reduction in the number of operations performed at the encoder for motion estimation by over two orders of magnitude while introducing minimal degradation to the decoded video compared with full search encoder-based motion estimation.

This paper describes a new compression algorithm, termed network-driven motion estimation (NDME), which reduces the power dissipation of wireless video devices in a networked environment by exploiting the predictability of object motion. Since the location of an object in the current frame can often be predicted accurately from its location in previous frames, it is possible to optimally partition the motion estimation computation between the portable devices and high powered compute servers on the wired network. In network-driven motion estimation, a remote high-powered resource at the base-station (or on the wired network), predicts the motion vectors of the current frame from the motion vectors of the previous frames. The base-station sends these predicted motion vectors to a portable video encoder, where motion compensation proceeds as usual. Network-driven motion estimation adaptively adjusts the coding algorithm based on the amount of motion in the sequence, using motion prediction to code portions of the video sequence which contain a large amount of motion and conditional replenishment to code portions of the sequence which contain little scene motion. This algorithm achieves a reduction in the number of operations performed at the encoder for motion estimation by over two orders of magnitude while introducing minimal degradation to the decoded video compared with full search encoder-based motion estimation.