Jun Ha

Energy-Aware Pure ALOHA for Wireless Sensor Networks*This work was supported by University IT Research Center Project.

A wireless sensor network is a network of compact micro-sensors equipped with wireless communication capability. In a wireless sensor network, saving energy is a critical issue. Furthermore, a sensor node is expected to face many difficulties in signaling and computing. As a MAC scheme for a wireless sensor network, we thus propose an energy-aware version of pure ALOHA scheme, where rather than sacrificing the simplicity of pure ALOHA, we take a straightforward approach in saving energy by trading off throughput performance. First, we add a step of deciding between stop and continuation prior to each delivery attempt for a MAC PDU. Secondly, we find an optimal stopping rule for such a decision in consideration of the losses reflecting energy consumption as well as throughput degradation. In particular, we note that the results of delivery attempts are hardly predictable in the environment that sensor nodes contend for the error-prone wireless resource. Thus, presuming that only partial information about such results is available to sensor nodes, we explicitly draw an optimal stopping rule. Finally, numerical examples are given to demonstrate the expected losses incurred by optimal stopping rules with full and partial information.

Occupancy regulation for reordering buffer at 3GPP's ARQ

3GPP RLC protocol specification adopted an error control scheme based on selective-repeat ARQ. In the 3GPPs ARQ, distinctive windows are installed at the transmitting and receiving stations so that each station is prohibited to send or receive a data PDU out of its window. As in a general selective-repeat ARQ, a classical problem of reordering is inherent in the 3GPPs ARQ. While a decrease in window size reduces the occupancy at the reordering buffer, such a decrease incurs degradation in delay and throughput performance. First, for the purpose of investigating the effect of window size on throughput performance, we develop an approximation method to calculate the maximum throughput of the 3GPPs ARQ. Secondly, aiming at reducing the occupancy at the reordering buffer as well as suppressing a deterioration of delay and throughput performance, we propose a scheme of regulating the occupancy at the reordering buffer as an additive to the 3GPPs ARQ. In the occupancy regulation scheme, a threshold is created in the receiving stations window and a data PDU out of the threshold (but within the window) is treated according to a go-back-N ARQ. For judging the effectiveness of the occupancy regulation scheme, we investigate peak occupancy, mean delay time and maximum throughput exhibited by the proposed scheme in various environments. From numerical examples, we confirm the validity of our approximation method and conclude that the occupancy regulation scheme is able to improve the delay and throughput performance of the 3GPPs ARQ without inflating the occupancy at the reordering buffer.

Stability of Game-Theoretic Energy-Aware MAC Scheme for Wireless Sensor Networks

A wireless sensor network is a network of compact micro-sensors equipped with wireless communication capability. In a wireless sensor network deployed to collect and update data, sensor nodes are usually battery-powered. Moreover, they are hardly able to support complex computation. Such unique features give rise to a critical issue; how to devise a simple energy-aware MAC scheme. Intending to resolve the issue, we consider a MAC scheme based on p-persistence slotted ALOHA. To determine the value of the attempt probability p, we construct the p-persistence slotted ALOHA as a simple non-cooperative game which involves generalized payoffs reflecting energy saving and throughput. Then, we derive a Nash equilibrium in a closed form. A Nash equilibrium is known to possibly incur a protocol failure and network instability. Using the Nash equilibrium, we thus investigate the effect of the payoffs on the stability of a wireless sensor network. For the stability of the wireless sensor network, numerical examples suggest an appropriate reflection of the traffic load on the payoffs.

Demand and grant MAC schemes for best effort service in wireless MAN

In the IEEE 802.16 Wireless MAN standard, the best effort service class is ranked on the lowest position in priority and is assisted by a MAC scheme based on reservation ALOHA. The details of the MAC scheme, however, are not specified in the standard. In this paper, we present the rules for deciding the amount of resource to demand as well as the rules for deciding the amount of resource to grant. Combining these rules, we then construct candidate MAC schemes for the best effort service. In designing a MAC scheme for the best effort service, the throughput and delay performance must be considered since only scarce resource may be available after resource is preferably granted to other service classes. For the performance evaluation, we thus develop a numerical method to approximately calculate the throughput in the saturated environment. Using the approximation and simulation methods, we then investigate the throughput and delay performance exhibited by each candidate MAC scheme. From the numerical results, we observe that the exhaustive demand rule paired with the full grant rule shows superior performance.

Uni- source and multi-source m -ary tree algorithms for best effort service in wireless MAN

IEEE 802.16 WirelessMAN standard specifies the air interface of broadband wireless access systems providing multiple services. In the wireless MAN, the best effort service class is ranked on the lowest position in priority and is assisted by a MAC scheme based on reservation ALOHA. In such a MAC scheme, a collision of resource requests is unavoidable so that wireless MAN standard adopted a truncated binary exponential backoff algorithm to arbitrate request attempts. However, it was revealed that a truncated binary exponential backoff algorithm may deteriorate delay and throughput performance due to its capture or starvation effect. Aiming at improving such performance, we propose uni-source and multi-source m-ary tree algorithms as alternatives to resolve request collisions in a wireless MAN. For the uni-source m-ary tree algorithm, we first develop an analytical method to calculate the maximum throughput. Secondly, using the analytical method as well as simulation method, we evaluate maximum throughput, mean and variance of MAC PDU delay. From numerical results, we confirm that proposed algorithms invoke superior delay and throughput performance to a truncated binary exponential backoff algorithm.

Unisource and Multisource Tree Schemes for Collision Resolution in Wireless MAN

IEEE 802.16 wireless MAN standard specifies the air interface of broadband wireless access systems providing multiple services. In the wireless MAN, the best effort service class is ranked on the lowest position in priority and is assisted by a MAC scheme based on reservation ALOHA. In such a MAC scheme, a collision of resource requests is unavoidable so that the wireless MAN standard adopted a truncated binary exponential back-off scheme to arbitrate request attempts. While an exponential back-off scheme is simple to implement, its capture or starvation effect was revealed to deteriorate the fairness in short-term throughput and delay variance in the long term. Aiming at improving the throughput and delay performance, we thus propose the unisource and multisource m-ary tree schemes as alternatives for resolving request collisions in a wireless MAN. For the unisource tree scheme, we first develop an analytical method to exactly calculate the throughput in the saturated environment. Using the analytical method and simulation method as well, we then evaluate the saturated throughput, mean of MAC PDU delay and variance of MAC PDU delay in each proposed scheme. From the numerical examples, we confirm that the unisource and multisource m-ary tree schemes invoke superior throughput and delay performance to a truncated binary exponential back-off scheme.

Supplement schemes to 3GGP ARQ for reducing occupancy and sojourn at re-ordering buffer

3GPP RLC protocol specification adopted an error control scheme based on window-controlled selective-repeat ARQ. As a member of selective-repeat clan, the 3GPP ARQ inherently possesses the problem of re-ordering at the receiving station. In this paper, aiming at suppressing the occupancy and sojourn time at the re-ordering buffer while maintaining the level of delay and throughput performance, we propose two schemes (identified as confined selective-repeat and extended go-back-N schemes) as a supplement to the 3GPP ARQ. For the performance evaluation of the proposed schemes, we first develop an approximation method to calculate the maximum throughput of a window-controlled ARQ. Secondly, using a simulation method, we investigate the peak occupancy, mean sojourn time, mean delay time, and maximum throughput exhibited by the schemes in various environments. From numerical examples, we confirm the validity of our approximation method and conclude that the proposed schemes are able to improve the delay and throughput performance of the 3GPP ARQ without inflating the occupancy and sojourn time at the re-ordering buffer.