Arvind Krishna

Enhancing throughput over wireless LANs using channel state dependent packet scheduling

Unlike wired networks, packets transmitted on wireless channels are often subject to burst errors which cause back to back packet losses. Most wireless LAN link layer protocols recover from packet losses by retransmitting lost segments. When the wireless channel is in a burst error state, most retransmission attempts fail thereby causing poor utilization of the wireless channel. Furthermore, in the event of multiple sessions sharing a wireless link, FIFO packet scheduling can cause the HOL blocking effect, resulting in unfair sharing of the bandwidth. This observation leads to a new class of packet dispatching methods which explicitly take the wireless channel characteristics into consideration in making packet dispatching decisions. We compare a variety of channel state dependent packet (CSDP) scheduling methods with a view towards enhancing the performance of the transport layer sessions. Our results indicate that by employing a CSDP scheduler at the wireless LAN device driver level, significant improvement in the channel utilization can be achieved in typical wireless LAN configurations.

On the capture probability for a large number of stations

The probability of capture under a model (for a land mobile radio direct sequence spread spectrum system) based on the ratio of the largest received power to the sum of interference powers is examined in the limit of a large number of transmitting stations. It is shown in great generality that the limit depends only on the capture ratio threshold and the roll-off exponent of the distribution of power received from a typical station. This exponent is insensitive to many typical channel effects such as Rician or Rayleigh fading and log-normal shadowing. The model is suitable for large systems with noncoherently combined interference.

On the randomization of transmitter power levels to increase throughput in multiple access radio systems

To enhance the throughput of a slotted random access protocol in a radio communication system, we describe the use of a scheme in which multiple power levels are used at the transmitters. We first consider a situation in which n transmitters are simultaneously trying to send a packet to a central receiving station using a time‐slotted access protocol, like slotted ALOHA. Each of these transmitters randomly chooses one of m discrete power levels during each attempt to send a packet. One of the simultaneously sent packets can often be successfully received due to the power capture effect. We consider two types of capture models: (1) one in which the transmitter with the largest received power captures the channel, and (2) one in which the transmitter captures the channel only if its signal‐to‐interference ratio is above some threshold when received at the central station. In this paper, we determine the optimal transmit probabilities for the power levels as well as the optimal values of the power levels themselves, when their range is constrained and for cases both with and without Rayleigh fading. After determining the precise optimal power levels and probabilities for maximizing the capture probabilities (i.e., for a given n), we propose a less complex, but nearly optimal, approximate approach based on using logarithmically equi‐spaced levels. After demonstrating the closeness of our suboptimal results to the optimal results, we apply our approach to the problem of optimizing the throughput of the slotted ALOHA protocol for a case in which the input traffic is generated according to a Poisson process. Several numerical examples are presented along with a demonstration of how the optimal choice of power levels and probabilities can enhance throughput relative to previous ad hoc methodologies.

Pseudocyclic maximum-distance-separable codes

The (n, k) pseudocyclic maximum-distance-separable (MDS) codes modulo (x/sup n/-a) over GF(q) are considered. Suppose that n is a divisor of q+1. If n is odd, pseudocyclic MDS codes exist for all k. However, if n is even, nontrivial pseudocyclic MDS codes exist for odd k (but not for even k) if a is a quadratic residue in GF(q), and they exist for even k (but not for odd k) if a is not a quadratic residue in GF(q). Also considered is the case when n is a divisor of q-1, and it is shown that pseudocyclic MDS codes exist if and only if the multiplicative order of a divides (q-1)/n, and that when this condition is satisfied, such codes exist for all k. If the condition is not satisfied, every pseudocyclic code of length n is the result of interleaving a shorter pseudocyclic code. >

Performance of shuffle-like switching networks with deflection

Four packet-switched networks using shuffle-exchange interconnections and deflection routing are analyzed. The first two are well-known networks based solely on shuffle interconnections, and the other two are variations in which the negative effects of deflection are reduced. Approximate state equations are given under a uniform traffic assumption. The equations predict the distribution of packet delay and can be used in situations where packets are assigned priorities. The four networks are briefly compared to each other and to Batcher-Banyan sorting networks and hypercube deflection networks.<<ETX>>

Analysis of a wireless MAC protocol with client-server traffic and capture

We analyze an efficient medium access control (MAC) protocol for use in a single cell of a wireless local area network (LAN). A fixed frame structure with two periods is used; one period for making reservations using a slotted Aloha protocol and a second period for data transmission. The MAC protocol operates in a centralized manner in which a single station, the base station, accepts reservations (transmission or data requests) that are made by the remote stations, and then schedules the times in which they will transmit or receive data. We consider two different types of client-server traffic models, an open-loop and a closed-loop model. In the open-loop model, a remote station can generate reservation requests for the base station whether or not it has received a response from the base station. In the closed-loop case, the remote station waits for a response before generating a new request. The performance of the MAC protocol was analyzed exactly for both the open and closed-loop traffic models and for cases in which transmission errors and a mean-value type of radio capture model are included. In addition, different policies were considered for the transmission strategy that is used by the remote stations when they attempt to make reservations. We derive exact results for the mean throughput and waiting times as well as for the queue length distributions. We also derive an approximate Markov chain to treat a case in which a fixed-position capture model is used. Several types of behavior are illustrated through the use of numerical examples. >

Design issues in wireless LANs

Wireless Local Area Networks (LANs), while similar in some respects to wired LANs, have many differences that impact their design. The key differences are the unreliable natnre of wireless communication and the mobility of stations enabled by their wireless capability. Various design alternatives for wireless LANs are discussed and the advantages and disadvantages of each choice are argued. The discussion is restricted to the design alternatives that are impacted by the wireless media and the mobile natnre of stations for the physical layer, Medium Access Control (MAC) protocol, topology choices for the connectivity of stations, and the networking layer. The choices that have been made in current systems are briefly described imd the potential advantages of alternative designs are discussed.

An architecture and implementation toward multiprotocol mobility

The challenge is to design a set of protocols that provide seamless and robust network connectivity to mobile users for a variety of existing network protocol suites. An important additional constraint is that the design should not require any changes to the existing stationary infrastructure and it should interoperate with the existing protocol stacks at the stationary hosts. We propose a mobile data link (MDL) architecture that endeavors to meet this design challenge. The architecture hides the topological effects of mobility at the data link layer, thus making topology changes due to mobility transparent at layers above the data link. This provides a single mechanism to support a variety of distributed applications that use different protocol stacks without requiring changes to the implementations of the stacks. Further, the-solution provides the enhanced degree of mobility provided by prior network layer solutions. The novel features of the proposed solution are the use of the network layer infrastructure for carrying data link traffic, the ability to dynamically maintain a data link overlay over a common network layer infrastructure, and efficient handling of the broadcast traffic. We have developed an MDL architecture that allows data link networks (formed by the interconnection of LAN segments by bridges) to be dynamic, thereby supporting mobile hosts, eliminating the need for manual configuration. Moreover, our scheme efficiently handles broadcasts and multicasts. >

System design issues for low-power, low-cost short range wireless networking

The emergence of battery powered handheld devices as popular computing devices is presenting new challenges. Among the most important challenges is the need to provide a low cost, low power, indoor wireless networking access to handheld devices. The constraints posed by battery power and cost require a careful re-evaluation of system design issues at all layers of the protocol stack. We present the design of a short range wireless networking system called BlueSky which is being developed at IBM Research to address these challenges. We show that the optimization objectives for short range indoor wireless systems are quite different from those for traditional cellular wireless systems. We also argue that in the next millennium the primary optimization criteria for the design of (short range) wireless systems will shift from the traditional spectral efficiency towards battery lifetime and cost.

Analytic performance modeling for a spectrum of multithreaded processor architectures

The throughput of pipelined processors suffers from delays associated with instruction dependencies and memory latencies. Multithreaded architectures attempt to hide such delays by sharing the processor with multiple instruction streams. In this paper we develop a comprehensive analytic framework to quantitatively evaluate the performance of a wide spectrum of mulithreaded machines, ranging from those that are capable of switching threads every cycle to those that switch threads only on long delays. The models are validated against previously published simulation and modeling results, and then used to assess the performance potential of multithreading given current processor technology.<<ETX>>