Research

Nanotechnology

Molecular Communication

This research explores the possibility of molecular communication as a solution for communication between nanomachines. Nanomachines are artificial or biological nano-scale devices that perform simple computation, sensing, or actuation. Molecular communication provides a mechanism for nanomachines to communicate over a short distance (adjacent nanomachines to tens of micrometers) using molecules as a communication carrier. Existing research focuses on understanding biological nanomachines and also on artificially creating counterparts of biological nanomachines. Very few research studies address communication aspects of nanomachines. Communicating nanomachines can spur the creation of entirely new applications such as a nano-scale distributed computing system or a nano-scale sensing system.

The class of molecular communication systems considered in this research consists of sender nanomachines, receiver nanomachines, carrier molecules, and the environment that these operate in. Senders and receivers include biological (such as cells) and biologically derived (such as molecular motors or sensors taken from biological systems) nanomachines that are capable of emitting and capturing carrier molecules (such as proteins, ions, or DNA). The environment is the aqueous solution that is typically found within and between cells. Please see the project page for more detail.

Researchers: Tadashi Nakano, Michael Moore, Ryota Egashira, Akihiro Enomoto

Middleware

The Bio-Networking Architecture

The near future networks, which will be orders of magnitude more complex and larger than current networks, should exhibit self-organization with inherent support for autonomy, mobility, scalability, adaptability to environmental changes in networks, and survivability/availability from massive failures and attacks. In order to address these requirements, we have been investigating the Bio-Networking Architecture, which is a new network application architecture that models network objects (resources) after biological concepts and mechanisms. The key target research areas include biologically-inspired adaptation mechanisms, emergent application composition mechanisms, distributed object (resource) discovery mechanisms, and middleware platform software. Please see the research activities page for more details.

Researchers: Jun Suzuki, Tadashi Nakano, Michael Moore, Keita Fujii, Ryota Egashira

Peer to Peer Discovery

We envision the future network where a variety of data and applications are highly distributed and a large number of users often join or leave. In such a network environment, future applications need to locate distributed network object (i.e., data, application, users) that meet a certain search criteria such as keywords. We develop discovery mechanisms that are decentralized and are based on concepts of keyword similarity between network objects, history of past discovery performance, and formation of keyword strengths based on user preference.  These concepts address issues of efficiency, scalability, adaptability, and usability that arise when performing discovery in a large-scale and dynamic network environment. Please see the research page for more details.

Researcher:  Michael Moore, Ryota Egashira

Dynamic Service Composition

Dynamic composition of complex services from primitive components brings flexibility and adaptability to future applications. By properly selecting and combining components on demand, applications would adapt to individual user preference and would consider available context information. This research is focusing on design and development of a component model, middleware and service composition mechanism in order to enable dynamic service composition. Please see the project page for more detail.

Researcher:  Keita Fujii, kfujii@ics.uci.edu

Sensor Network

A mobile sensor network is composed of a distributed collection of nodes,each of which has sensing, computation, communication and locomotion capabilities. Locomotion facilitates a number of useful network capabilities, including the abilities to self-deploy and self re-organize; that is, starting from some random initial configuration, the nodes in the network can coordinate to achieve certain coverage according to application's requirement and their local environment.. Also the nodes can manage to recover the coverage after some network dynamics. e.g. sensor failure, network partition... In this project, we propose an algorithm to allow mobile sensors autonously form barrier and sweep coverage around certain mobile target(s) to minimize the undetected penetration through the barrier. The algorithm applies the Lennard-Jones potential field theory by modeling mobile sensors and targets as different virtual molecues and employing virtual forces between them. The algorithm works in a fully distributed and self-organizing way.

Researchers:  Mei Yang, meiy@ics.uci.edu

Light-weighted Data Storage and Retrieve Scheme in Wireless Sensor Networks

With the advancements in micro electronic technology, it is possible to deploy a wireless network with a large number of small and light-weight sensors for monitoring applications. Such a sensor network not only carries the data traffic (i.e., sensor data traffic), but also processes and stores the sensor data within a network for users to later retrieve the sensor data of interests. Thus, it is required to device a scheme that is scalable to a large number of deployed sensors and to a large amount of sensor data in the network, adaptive to dynamically changing user interests, under the constraints of highly limited sensor processing and memory resources on easy-to-fail small sensors. We propose to develop a scalable, robust and adaptive data retrieval scheme for a sensor network with a large number of deployed sensors. The proposed scheme applies DHT (Distributed Hash Table) technique to the attributes of sensor data to generate storage locations (i.e. a track) for sensor data. The proposed scheme differs from other existing research by dynamically choosing the most popular attributes to create tracks storing the data sets with attributes satisfying the popular user queries, when changes of user interests are represented by different attributes in user queries.

Researchers:  Yi Pan, ypan@ics.uci.edu, Jun Lu, lujun@ics.uci.edu , Satoshi Yamamoto

Sensor Network Research

One important measurement of the performance of a sensor network is how well the sensor network can monitor a covered area, namely network coverage. In this project, we propose a scheme to improve the network coverage of a sensor network. A probabilistic model of network coverage is proposed, and the concept of Sensing Denomination (SD) of a sensor is introduced to measure the contribution of the sensor to the network coverage. Based on the location information of neighboring sensors, each sensor in a sensor network can calculate its SD value in a distributed way. The sensors then probabilistically schedule their active and hibernating states according to their SD values so that sensors with higher SD value have more chance to stay active. Through this self-scheduling process, a sensor network can provide better coverage given certain energy consumed.

Researchers:  Jun Lu, lujun@ics.uci.edu

Traditional Networking

Smooth Handoff for Stream Media

The current mobile wireless networks often fail to provide smooth handoffs due to the following two reasons: 1) packets may get lost due to the re-routing caused by handoffs; 2) due to the disparity in available bandwidth among wireless cells, handoffs may cause congestion. To avoid stream disruption and drastic quality degradation of stream transmission during handoffs, we developed a multi-path handoff scheme to solve that problem. In the proposed scheme, multiple paths are set up during the handoff. Transmission rate on each path is controlled separately to avoid congestion. By transmitting important data (base layer) in video stream redundantly through multiple paths, we avoid stream disruption during the handoff; by exploiting extra bandwidth on high bandwidth path to send additional data (enhancement layer), we improve the video quality. Simulation is conducted to evaluate the performance of the proposed scheme. 

Researchers:  Yi Pan, ypan@ics.uci.edu

QoS in High-speed Networks

Next generation high-speed IP networks are expected to support diverse multimedia applications in a scalable manner.   In such networks, it is important to provide quality-of-service (QoS) guarantees to a wide range of applications in a scalable manner. We propose and study a new QoS framework, called On-Demand QoS Path (ODP), which provides end-to-end QoS guarantees to individual flows with much less overhead than IntServ, while keeping the scalability characteristic of DiffServ.

Researchers:  Yan Huang, yanh@ics.uci.edu, Mei Yang, meiy@ics.uci.edu

Inter-Domain Resource Management

The inter-domain resource exchange (iREX) project is about enabling domains to exchange premium traffic in a bi-lateral manner that will comprise of all participating domains advertising a price for their available bandwidth and the domain needing services contacting these domains individually and setting up a path based on the sending domain's own criteria.

Researchers:  Ariffin Datuk Yahaya, ariffin@ics.uci.edu

Applying Wavelet-Denoising in Congestion Control with Hybrid Internet Traffic

In the current Internet, congestion control is performed in a TCP/AQM model. At the end Systems, Internet traffic is mainly controlled by TCP protocols, which apply an additive increasing and multiplicative decreasing (AIMD) congestion control scheme. To avoid oscillation of queuing length at the bottleneck links and provide early congestion indication to the end systems, Active Queue Management (AQM) controllers are applied at the bottleneck routers, e.g. RED is the de facto standard technique used in the routers. However, the control parameters of AQM controllers are hard to set up in the TCP/AQM model, since they are highly sensitive to traffic load and other network parameters, such as round trip time, link capacity, etc. In addition, unresponsive traffic, such as HTTP traffic, is not responsive to AQM congestion control signals (i.e. packet loss/mark probability) and may further affect the TCP/AQM congestion control model by introducing false congestion signals to the end systems. Designing a robust TCP/AQM controller with the presence of a hybrid of unresponsive and responsive traffic remains a challenging problem. Our design is the first to introduce a wavelet-based traffic denoiser to remove the impact of unresponsive traffic while still being able to freely choose the optimal parameters for the responsive traffic flows. This enables a new paradigm in the design of AQM controller with hybrid traffic in the Internet.

Researchers:  Yi Pan, ypan@ics.uci.edu

Ad Hoc Network Research

Existingresearch work has provided the multi-rate capability in MAC layer for ad hoc networks. But their work didn’t investigate how this multi-rate capability will impact on the performance of multi-hop ad hoc networks. We will investigate how we can employ this multi-rate capability in routing layer. We believe that the throughput of the network will be improved by wisely distributing the traffic in the network using routing control. Our approach is to take the multi-rate information into consideration of the routing decision. We propose to use average transmission delay to carry the multi-rate information. Then the routing decision is based on the smallest delay. We implement the scheme by modifying the existing protocols including IEEE 802.11 and DSR.

Researchers:  Weilin Zeng, wzeng@ics.uci.edu

Previous Research Projects