Soonmok Kwon

Dynamic timeout for data aggregation in wireless sensor networks

In-network data aggregation is a cross-layer routing technique which is essential for wireless sensor networks with a large number of nodes. Although the technique enhances energy efficiency by reducing packet transmissions, it may require routing-layer waiting time which introduces additional delays in packet delivery. This implies that there exists an energy-delay tradeoff which can be controlled depending on the delay level an application can tolerate. In this work, we first perform an in-depth analysis of the data aggregation process and validate the strong relationship between the amount of currently accumulated data in a node and the data aggregation timeout to achieve energy efficiency with little delay overhead. Then, we propose a novel timeout control scheme which dynamically changes timeout according to the amount of currently accumulated data in a node. Our scheme is based on local information without involving control messages so that it is not only robust but also light-weight. Evaluation results based on an implementation of the wireless sensor network show that our scheme facilitates efficient tradeoff between energy and delay. Thus, high energy efficiency can be achieved with little overhead in delay, and vice versa.

An Efficient Tree Structure for Delay Sensitive Data Gathering in Wireless Sensor Networks

It is important to design an energy efficient data gathering tree structure for wireless sensor networks. As for the energy efficiency, networks overall energy consumption and per-node fairness have been studied. Note that the minimum degree spanning tree (MDST) is optimal in the sense of per-node fairness. We believe that the per-node fairness is of practical interest and that the delay bound associated with it must be investigated. Unfortunately, no such efforts have been made so far. In this paper, we propose a tree structure, called the DB- MDST, which modifies MDST to reduce the height and extend its degree. With the DB-MDST, the near optimal delay bound (i.e., comparable to the one obtained using the shortest path tree) can be achieved at the cost of as much energy as consumed by the MDST while it cannot be achieved with the MDST. Simulation results support the claim stated.

Distributed and localized construction of routing structure for sensor data gathering

Several routing structures have been proposed for data gathering in a wireless sensor network. They are considered to be near-optimal with respect to energy efficiency or delivery delay, but they overlook the construction cost of a routing tree which may make trouble in implementation. Our primary goal in this work is to construct a routing tree with negligible cost, which performs as well as those near-optimal schemes. First, we propose a distributed and localized framework for tree construction called Local Parent Designation (LPD), where the node status information is exchanged locally and decision based on the local information leads us to construct a routing tree. Secondly, we extend LPD to so called LPD-Local Fix (LPD-LF) to reduce further the construction cost with less local information exchange. Simulation results validate that our goal is achieved.

A polling method using orthogonal signalling for Wireless Local Area Networks

The point coordination function (PCF) of the IEEE 802.11 WLAN facilitates deterministic frame transfer by a polling mechanism. This enables to support quality of service. Despite this benefit the PCF could be inefficient at low-to-medium traffic loads because it always polls all stations regardless of their actual demand. To overcome this drawback, we introduce a novel polling scheme where all stations simultaneously report their status by sending orthogonal signals, and a BS identifies and polls only those stations having data to send. The feasibility of orthogonal signalling is evaluated in a realistic channel model where multipath fading and additive white Gaussian noise are applied. Moreover, channel utilization of the proposed scheme is investigated analytically, and compared with other schemes through simulation. The evaluation results show that the proposed scheme considerably outperforms other schemes at low-to-medium traffic loads.

Practical approach to sensor data gathering

Various routing structures have been proposed for the data gathering in wireless sensor networks, which are supposed to be efficient in terms of energy consumption and delivery delay. Unfortunately, most of these structures assume not only a static and stable network environment but also the network-wide information such as global topology. These assumptions lead to intolerable overhead in the construction of a routing structure in many practical scenarios. We present a novel solution termed as Local Parent Designation (LPD), which is mostly based on local information such that each node locally designates its parent in the routing tree. The resultant tree is characterized by each nodepsilas local decision on its eligibility to be a parent and priority of designating a parent. While using such minimal resources, LPD is robust to network changes and flexible to various application demands. Moreover, simulation results show that the performance of LPD is comparable to that of other high-cost schemes in terms of energy and delay.

Orthogonal signaling-based queue status investigation method in IEEE 802.11

IEEE 802.11 specifies four different medium access control (MAC) protocols to coordinate multiple access in a wireless local area network (WLAN). Since several tens of stations can operate in a WLAN, the performance of MAC protocols is important for overall network efficiency. It has been observed that the IEEE 802.11 MAC protocols can be improved by knowing which station has a non-empty queue, i.e., queue status. The point coordination function (PCF) can use this information to avoid polling a station that has no pending data. The HCF controlled channel access can adapt polling parameters based on queue status information, especially when scheduling a bursty and variable bit-rate traffic. Previously suggested methods are rather limited in terms of accuracy and efficiency. In this paper, we propose a novel method to investigate the queue status of multiple stations by exploiting orthogonal signaling. With synchronous transmission of orthogonal codes and symbol level signal processing, the method allows all of the associated stations to report their queue status at the same time. Challenges that can arise in the implementation of the proposed method are identified, and their solutions are suggested. The feasibility of detecting orthogonal signals is thoroughly tested on a realistic channel model. To demonstrate the performance improvement of a MAC protocol, we applied the proposed method to PCF. Both analysis and simulation show that the modified PCF significantly outperforms not only the original PCF but also other previously suggested PCF enhancements.

Timeout control for data aggregation in sensor networks with packet size limit

In-network data aggregation is one of the key technologies in wireless sensor networks. Because data aggregation inherently incurs a tradeoff between energy efficiency and delivery delay, several solutions have been proposed to control the waiting time before transmission of an aggregated data (i.e. data aggregation timeout). In this work we consider the case when one packet can carry multiple sensing data elements and this aggregation is limited to a finite number due to the limitation of packet size. Based on the idea that the timeout may be determined by the current number of waiting data elements, we propose a scheme for varying timeout, analyze its performance using a stochastic model, and show how to customize the varying timeout in the proposed scheme for energy efficiency or low delay. Experiment results show that our analytical model accords well with the real world implementation and our scheme facilitates efficient tradeoff between energy and delay, that is, high energy efficiency can be achieved with little overhead in delay, and vice versa.

Orthogonal Signaling-based Sensing Data Reporting for Cooperative Spectrum Sensing in Cognitive Radio

Cognitive radio (CR) features opportunistic access to spectrum when licensed users (LU) are not operating. To avoid interference to LU, cognitive users (CU) need to perform spectrum sensing. Because of local shadowing, fading, or limited sensing capability, it is suggested that multiple CUs cooperate to detect LU. In cooperative spectrum sensing, CUs should exchange their sensing data with minimum bandwidth and delay. In this paper, we introduce a novel method to efficiently report sensing data to the central node in an infrastructured OFDM-based CR network. All CUs simultaneously report their sensing data over unique and orthogonal signals on locally available subcarriers. By detecting the signals, the central node can determine subcarrier availability for each CU. Implementation challenges are identified and then their solutions are suggested. The proposed method is evaluated through simulation on a realistic channel model. The results show that the proposed method is feasible and efficient.