Bogdan Ciubotaru

Subjective Assessment of Region of Interest-Aware Adaptive Multimedia Streaming Quality

Adaptive multimedia streaming relies on adjusting the video contents bit-rate meet network conditions in the quest to reduce packet loss and resulting video quality degradations. Current multimedia adaptation schemes uniformly adjust the compression over the entire image area. However, research has shown that user attention is focused mostly on certain image areas, denoted areas of maximum user interest (AMUI), and their interest decreases with the increase in distance to the AMUI. The region of interest-based adaptive multimedia streaming scheme (ROIAS) is introduced to perform bit-rate adaptation to network conditions by adjusting video quality relative to the AMUI location. This paper also extends ROIAS to support multiple areas of maximum user interest within the same video frame. This paper presents the performance analysis of ROIAS in terms of the impact on user perceived video quality measured using subjective video quality assessment techniques based on human subjects. The tests use a wide range of video clips, which differ in terms of spatial and temporal complexity and region of interest location and variation. A comparative evaluation of both subjective and objective video quality test results is performed and demonstrate the benefit of using ROIAS for adaptive video quality delivery.

Embedding Data in Video Stream using Steganography

The transmission of streaming data, because of its real-time nature, is subject to bandwidth constraints. We propose a MPEG-2 compliant decoder which uses data hiding to transmit additional data with no extra bandwidth while creating an almost imperceptible degradation of the picture. The additional data can be anything from subtitles to correction signals and watermarking.

Digital Watermarking using Karhunen-Loeve transform

Digital watermarking is a very common technique used against illicit copying and distribution of copyright material. This paper presents improved methods for information hiding. We propose a new approach, based on the decorrelation property of the Karhunen-Loeve transform (KLT).

A Slow-sTart Exponential and Linear Algorithm for energy saving in wireless networks

Limited battery capacity seriously restricts the usage of wireless devices in wireless communication networks. Power saving MAC layer protocols put wireless network interface into sleep mode in order to conserve energy. However, increased energy savings may sacrifice delivery performance and consequently the user perceived Quality of Service (QoS). This paper introduces Slow-sTart Exponential and Linear Algorithm (STELA), a novel battery saving mechanism which saves up to 70% energy compared with the algorithms employed by IEEE 802.11 while preventing the delivery performance from degrading significantly and consequently reducing the negative impact on user Quality of Experience (QoE).

Emergent Behavioral Modeling Language in Obstacle Avoidance

Artificial distributed systems employ sophisticated algorithms that are well matured and efficient but require a high degree of computational power. Opposite to artificial distributed systems, natural systems display emergent behavior by employing extremely simple algorithms that are very efficient when dealing with large number of entities but are unpredictable and nondeterministic. The formal modeling of such emergent behavior is presented in this paper by the introduction of a formal language called eBML. A simple yet effective obstacle avoidance algorithm is proposed and discussed here, based on studies of ant colonies which are modeled using eBML.

Vehicular-Communications-Based Speed Advisory System for Electric Bicycles

Smart transportation, which is an important dimension of smart cities, includes both intelligent and “green” transportation solutions. Cycling, as one of the most sustainable forms of transportation, is and should be an important component of smart cities. Electric bicycles, which are the most popular electric vehicles, subscribe to this type of transportation. They have several advantages when compared with traditional bicycles but have issues that relate to battery-limited capacity and long periods of charging as well. Consequently, energy-efficient solutions for electric bicycles are of very high research interest. Research on vehicular-communications-based energy-efficient solutions for electric vehicles is still in the early stages. Among electric vehicles, electric bicycles distinguish themselves as a special class as they have different characteristics and road-related requirements. This paper proposes a novel vehicular-communications-based speed advisory system for electric bicycles. The solution recommends strategic riding (i.e., the appropriate speed) when bicycles are approaching a signaled intersection to avoid high-power-consumption scenarios. The proposed approach includes a fuzzy-logic-based wind-aware speed adaptation policy, as, among all the other vehicles, bicycles are mostly affected by wind. Experimental results based on a real testbed and extensive simulation-based testing demonstrate that, by using the proposed solution, significant energy savings are recorded. In addition, an analysis on comfort-related metrics shows that the proposed solution can also contribute to improving the cycling experience.

Smooth Adaptive Soft Handover Algorithm for Multimedia Streaming over Wireless Networks

Inter-network mobility is achieved by allowing a mobile node to change its point of attachment to the network while preserving connectivity to its corresponding nodes. Most handover solutions proposed in the literature directly change the whole data flow from one network to another relying on only one network to transfer the entire data stream. These solutions involve a certain amount of quality degradation due to increasing loss and delay and suffer in terms of scalability, efficient resource allocation and resilience to different mobile node speeds. This paper proposes the Smooth Adaptive Soft-Handover Algorithm (SASHA) which increases the quality of the multimedia delivery process when performing handover in heterogeneous wireless environment by gracefully transferring the load from one connection to the other.

eBML: A Formal Language for Behavior Modeling and Application Development in Robotic Collectives

The new trend of miniaturizing the electronic devices leads to the necessity of small, relatively independent entities which can interact with each other and with other entities forming collective, associative environments. This paper describes the formal specification of the eBML modeling language and its use in programming intelligent terminals like robots or intelligent sensor nodes which compose such collaborative robotic environments.

Q-PASTE: A cross-layer power saving solution for wireless data transmission

This paper introduces a novel quality-oriented cross-layer energy-efficient data communication solution (Q-PASTE) for mobile devices, with two components at different network layers. First component, the Packet/ApplicaTion manager (PAT) is deployed at the application layer of both service gateway and client host. The gateway level PAT shapes traffic into bursts to reduce the wireless transceivers duty cycle. The client-side PAT monitors each active session and informs the Medium Access Control (MAC) layer about their traffic-related behavior. The second component, the Slow sTart Exponential and Linear Algorithm (STELA), deployed at MAC layer, adaptively adjusts the sleep/wake-up behavior of mobile device wireless interfaces in order to reduce energy consumption while also maintaining high Quality of Service (QoS) levels. Both mathematical analysis and simulation-based modeling and testing have been performed to evaluate Q-PASTE performance. Comparative results have shown significant energy saving achieved when using Q-PASTE, without compromising content delivery performance.

Connectivity Improvement in Wireless Sensor Networks Based on Mobile Nodes

In this paper we address the topic of wireless sensor networks deployment, redeployment and post-deployment network maintenance based on mobile robotic nodes in order to achieve optimal connectivity and minimum energy consumption. The research is based on the CORE-TX platform, which provides the necessary hardware and software modules used for evaluation and testing of the solutions discussed. The quality of the communication link has an important role in energy saving due to packet loss ratio and the required power level of the transceiver. The paper proposes and discusses an automated approach to network deployment and network maintenance based on local connectivity evaluation.