TCP-Friendly Rate Control (TFRC) has been widely adopted for advanced streaming applications over wired-wireless networks. TFRC applies an equation-based rate control scheme to provide smooth sending rate and perceptual quality in streaming applications. However, TFRC tends to malfunction in wireless environment if packet lost event was introduced by poor channel quality but network congestion. Therefore, TFRC cannot provide high quality-of-service for streaming applications over wired-wireless networks. In this paper, we proposed a novel one-way delay jitter based TFRC with end-to-end semantic over wired-wireless networks. This scheme not only provide smooth sending rate and TCP-friendly characteristics like standard TFRC, but also considerable increase the throughput by accurately estimating the available bandwidth in wired-wireless networks with bursty nature of background traffic. Simulation results show that our scheme conducts performance improvement without intrusiveness issue and even if background traffic is bursty over wired-wireless networks.
For the better security and safety of community, huge amount of video surveillance cameras are installed everywhere nowadays. It is thus highly desired to demand the security systems to real-time monitor and analyze events so as to generate real-time alerts and take appropriate real-time actions to minimize the damages caused by crimes and threats. Thanks to the success of wireless broadband access technologies, such as Wi-Fi and WiMAX/LTE technologies, real-time video surveillance and networking for large scale camera networks is now made possible. The applications include large scale community surveillance, environmental monitoring, military maneuvering, etc.In this talk, I will first talk about how a smart camera can effectively perform information exchange in a distributed manner of the processed results to neighboring cameras wirelessly to ensure effective camera calibration and multiple moving objects tracking across different cameras. By slight modification of the CSMA/CA access mechanism, we propose an idle probability based broadcasting method which adaptively adjusts the broadcast probability to improve broadcast effectiveness in a dense saturated camera network. For those event-detected cameras, the suspicious videos need to be uplinked to cloud servers for archiving or real-time dissemination. We propose a systematic cross-layer scheme, which incorporates an opportunistic scheduler and a QoE-driven resource allocation policy for OFDMA based networks, to appropriately adapt the video encoding to the assigned target bit rate from the basestation so as to ensure the efficient use of the limited bandwidth resources.
In views of the shortage of the wireless spectrum, the US Communications Commission (FCC) has adopted the rules that allow unlicensed devices to operate in TV bands so as to achieve better spectrum efficiency. To enable this new way of wireless communication, new technologies such as resource discovery and multi-channel resource allocation must be developed. In this paper, a multi-channel testbed is developed to demonstrate the feasibility of real-time spectrum sensing and multi-channel transmission for next-generation DSA networks.
In this paper, we investigate and propose a latency minimized CSMA variation to enhance the performance of event-driven wireless ad hoc network applications. By first analyzing latency performance based on p-Persistence CSMA, we show that the optimal transmission probability for latency minimized p-Persistent CSMA (LMPP) in an event-driven scenario equals to the optimal transmission probability for throughput maximization in a saturated wireless local area network when the number of competing nodes is much larger than the number of messages a node intends to receive and the packet length is a lot longer than the slot time. The proposed Latency Minimized Probabilistic CSMA/CA (LaMP) is a multiple access scheme with constant contention window size built atop standard 802.11 CSMA/CA, to approach the performance of LMPP. To deal with the critical contender estimation problem in LaMP, we also propose an Exponentially Decreased Matching (EDeM) strategy to approximate the number of contenders. Simulation results show that the proposed method consistently outperforms existing methods in latency performance.
This paper presents the necessity of application-based scheduling to maintain QoS requirements of all active applications in an indoor femtocell. A novel scheduler is proposed which takes into account the quality of service (QoS) requirements of each active application as well as whether their sustainability is at risk at a given time. The scheduler aims to maintain an efficient use of the wireless radio resources and strives to achieve a user and application specific notion of fairness of service reception. Unlike current solutions, the scheduler differentiates between not only different users but also between applications of each user, and schedules data transmission amongst these applications so that QoS requirements for all applications are satisfied.
In this paper, we introduce our own implementation of Map Reduce graph-theoretic algorithms for Email social network analysis on the Hadoop platform. Graph theory is a powerful tool for social network analysis and Map Reduce is a well-known paradigm for distributed parallel computing. However, based on our own experience, unlike writing conventional Java/C++ programs, writing Java programs to implement Map Reduce graph-theoretic algorithms is not straight-forward, even for some fundamental graph-theoretic algorithms. In this paper, for the problem of Email social network analysis, we compare the performance of cloud computing programs with that of conventional computer programs. We show that as long as the size of the input data exceeds a threshold, the cloud computing programs outperform their conventional counterparts.
In the past decade, people have witnessed more and more mass natural or man-made disasters. An efficient and flexible evacuation and rescue system is needed to save lives from various mass disasters. Although there exist some public emergency evacuation systems, these systems are not flexible to handle various disasters can not cooperate well. In addition, they are not efficient enough timely to update evacuation plans or routes to victims. In this work, we propose a framework for disaster relief, called Personal Evacuation And Rescue (PEAR) System. The PEAR system meets all the requirements of being an ideal emergency evacuation system. It is not only an effective alert information dissemination system, but also a pervasive and efficient emergency evacuation system. PEAR is a programable system whose multimedia contents can be always up-to-date. By utilizing wireless technologies and smart mobile devices, PEAR can save more people by providing most intuitive, timely, localized and cooperative evacuation routes for either indoor or outdoor evacuation.
"Skymesh" is an ad hoc wireless network using balloon to provide a temporary network for disaster recovery. The swing of the balloons causes instantaneous link disconnection and link quality degradation in skymesh. The resulting network quality significantly affects application-level quality such as video quality in remote monitoring using camera on the balloons. In this paper, a link-by-link video transmission scheme is proposed to realize more stable video quality than the conventional end-to-end video transmission scheme. In the proposed scheme, TCP connection is setup over each link long the path between the source and destination with video frame storing and retransmission capabilities of the orwarding nodes along the path. The performance of the ink-by-link scheme and end-to-end scheme are compared by emulation under random packet loss and burst packet loss model. It is shown that the link-by-link scheme improves throughput and video quality at the cost of delay. Trial and skymesh experiments are also carried out in Niigata University Campus. These experimental results confirm that the link-by-link scheme is effective for the end-to-end communication quality and will be also able to improve video quality of the remote monitoring system in skymesh.
Enterprises nowadays are approaching towards an all IP paradigm by subscribing to different access technology links from several service providers for reliability and redundancy while providing good Quality of Service (QoS). Global System for Mobile Communications (GSM) to Voice over IP (GSM-2-VoIP) convergence and vice-versa is a step forward towards this unification goal. It requires dynamic routing between IP, digital, analog and GSM networks. A decision-making framework is presented that can handle both the dynamic routing decision computation at private-public network border for inbound/outbound calls (GSM-2-VoIP) and the horizontal handover decision-making among various GSM cells. The decision engine takes multiple criteria into account while computing the routing and/or handover decisions (attributes from context of the request, platform's latest conditional parameters, business objectives of the company etc.) and/or (mobility management issues, user profile, traffic load etc.) respectively. An all open source real time platform consisting of Universal Serial Radio Peripheral (USRP), GNU Radio, an extended GSM stack (OpenBTS), OpenSIPS and Asterisk is developed to validate the proof of the concept. The system gives higher throughput and lower call dropping probability with an add-on susceptible delay offering good QoS and lower cost.
A cross-layer framework of interplay between multiple description coding (MDC) at the source and layered modulation using superposition coding (SPC) at the channel has recently been proposed as an effective approach to mitigate the vicious effects due to multi-user diversity and fast Rayleigh fading in wireless video broadcast/multicast. However, there has not been a rigorous and general end-to-end distortion analysis to determine if and when such interplay between the protection code (such as MDC) in the upper layer and the multi-resolution modulation by way of SPC in PHY/MAC layer for transmitting successively refined information can lead to a better end-to-end distortion performance than other possibilities. The paper provides a comprehensive information theoretical analysis on the end-to-end distortion. Mathematical proofs are derived to confirm the advantages and conditions of such cross-layer systems to always give the better distortion performance. Numerical analysis is conducted to verify the mathematical conclusions while unveiling some interesting observations.
Scheduling algorithms in WiMAX are of great importance due to their effect in delivering a high speed broadband while satisfying the traffic Quality of Service (QoS) constraints. A well designed scheduling algorithm should carefully deal with throughput maximization, delay constraints satisfaction and maintaining fairness among subscribers. This paper focuses on uplink (UL) scheduling effect on the performance of real time Polling Services (rtPS) QoS class performance using OFDM as a physical layer where rtPS is a service class for variable bit rate (VBR) data such as MPEG compressed video. Ensuring QoS constraints for rtPS class is challenging due to critical delay constraints and throughput requirements. The paper proposes an uplink scheduling algorithm called the Instantaneously Replacing Algorithm (IRA). The algorithm mainly schedules connections based on their Signal to Noise Ratio (SNR) but instantaneously replaces high SNR connections with connections that may violate their intended QoS requirements. The proposed algorithm is analyzed in the paper using an NS-2 simulation model. Comparing to another set of UL schedulers, simulation results show that the proposed algorithm enhances QoS Satisfaction in the network as it tends to minimize the delay and at the same time distribute the network resources in a fair manner among Subscriber Stations (SSs) while maintaining a throughput comparable to that achieved using SNR based approaches.
In wireless networks, bandwidth is relatively scarce, especially for supporting on-demand media streaming. In wired networks, multicast stream merging is a well-known technique for scalable on-demand streaming. Also, caching proxies are widely used on the Internet to offload servers and reduce network traffic. This paper uses simulation to examine a caching hierarchy for wireless streaming video distribution, in combination with multicast stream merging. The main purpose is to gain insight into the filtering effects caused by caching and merging. Using request frequencies, entropy, and inter-reference times as metrics, we illustrate how merging, caching, and traffic aggregation affect the traffic characteristics at each level. The simulation results provide useful insights into caching performance in a video streaming hierarchy.
Nowadays, users are reachable through a number of telecommunication devices, network operators, as well as services and applications. The calling users if provided with the appropriate information can manually select the most suitable way of communication. Although this capability already affects existing business models, the lack of a transparent and a dynamic solution renders it hard to use. To tackle this issue we suggest minor extensions in the SIP protocol as well as the use of middleware to allow calling users to dynamically select the best way to route their requests towards called users. These extensions are described in detail while execution examples are also provided.
The exploration of virtual environments in wireless mobile media devices has attracted the attention of researchers and developers mainly due to its potential applications in a variety of areas including entertainment, training, security, e-learning, etc. However, current technology of mobile devices lack the proper resources to handle complex and realistic 3D virtual environments. There has been a number of proposed ideas to solve this issue. The existing approaches use techniques that either employ limited user navigation modes or do not perform satisfactorily for interactive applications. In this paper, we propose a new protocol that offers the user a richer navigation by pre-streaming the necessary imagery data to generate new views as the user wanders within the 3D environment. We introduce the idea of key partial panoramas, i.e., panorama segments that cover movements in any direction by simply strafing from an appropriate key partial panorama and streaming the amount of lost pixels. We have implemented our ideas and evaluated it against two well-known approaches. Experimental results show that our solution outperforms the selected approaches by minimizing the delay between image updates and by allowing a more complex navigation scheme than previous works.
H.264 is the most recent standard for video compression, achieving not only the highest compression efficiency but also providing network friendliness and error resiliency. Data partitioning is one of the very important error resilience features of H.264. With data partitioning, each video slice is encoded into three different units of data with different importance. The encoded partitions containing the most important information should be protected against transmission error to ensure good picture quality. By virtue of multistreaming and partial reliability property of Stream Control Transmission Protocol (SCTP), we can set different priority or reliability levels for different data partitions. In this article, we investigate the impact of the loss of different partitions on picture quality. We present a comparative study of the possible solutions for transmission of H.264 video using SCTP, considering both partitioned and non-partitioned H.264 video. We demonstrate how reliability features of SCTP can be efficiently mapped to the error resiliency features of H.264 video.
In wireless sensor networks, bandwidth is one of precious resources to multimedia applications. To get more bandwidth, multipath routing is one appropriate solution provided that inter-path interferences are minimized. In this paper, we address the problem of interfering paths in the context of wireless multimedia sensor networks and consider both intra-session as well as inter-session interferences. Our main objective is to provide necessary bandwidth to multimedia applications through non-interfering paths while increasing the network lifetime. To do so, we adopt an incremental approach where for a given session, only one path is built at once. Additional paths are built when requi red, typically in case of congestion or bandwidth shortage. Interference awareness and energy saving are achieved by switching a subset of sensor nodes in a passive state in which they do not take part in the routing process. Despite the routing overhead introduced by the incremental approach we adopt, our simulations show that this can be compensated by the overall achieved throughput and the amount of consumed energy per correctly received packet especially for relatively long sessions such as multimedia ones. This is mainly due to the fact that a small number of non-interfering paths allows for better performances than a large number of interfering ones.
Video streaming is going to drive the Internet to a new height. However, how to distribute the high-quality video streams that are already delivered to the doorsteps of ordinary customers to almost all rooms in their dwellings still remains a technique challenge. There are various proposals for wired, ``no-new-wires" or wireless approaches, but their performance in real environment is yet to be examined. In this paper, we evaluate the performance of high-quality video streaming over short-range wireless networks, specifically Ultra-Wide Band (UWB), for a two-tiered broadband home network. Our analysis and experiment results reveal the unique features of UWB reservation schemes and their intrinsic throughput-latency tradeoffs, which indicates the efficacy of UWB for high-speed, in-room wireless access with both quality and mobility provisioning.
Reference Picture Selection (RPS) is a powerful error control technique for video streaming, but it consumes significant computational resources at streaming time. In this paper we present a two-stage H.264/AVC-compatible video encoder that enables the use of RPS at significantly reduced computational cost, and with only a minor penalty in coding performance. Further, we show how the compression performance gap between our two-stage encoder and the conventional H.264/AVC encoder can be eliminated by using additional disk space at the server.
This work presents the design of real-time applications for wireless sensor networks (WSNs) by using an algorithm based on data stream to process the sensor data. The proposed algorithm is based on sampling techniques applied to data histograms created from original data streams acquired by sensor nodes. As a result, the algorithm provides a sample of log n items to represent the original data stream of n elements. In this work, we show how to use the data reduction algorithm in real-time WSN design.
