Search

In future wireless ad hoc networks, transactions are expected to become one of the primary types of flows. Transactions require only a small number of packets to complete. Hence, providing optimal (shortest path) routes to such transactions consumes more energy than the actual data transfer. Conventional shortest path routing protocols are, thus, unsuitable for routing transactions. In this paper, we present a novel architecture, called TRANSFER, for transactions routing in large-scale wireless ad hoc networks. In our approach, we aim to reduce the total energy consumption by transactions as opposed to finding shortest path routes. Our architecture uses a hybrid approach, in which each mobile node obtains information about nodes in its proximity (zone), up to R hops away, using a proactive link-state protocol. Beyond the proximity, we introduce the novel notion of contacts that act as shortcuts to reduce the degrees of separation between the source of the transaction and the destination. We propose an efficient on-demand protocol for contact selection that does not assume knowledge of location information. Contacts are used during transactions and queries to discover valid routes in an energy-efficient manner. Extensive simulations are used to evaluate the performance of our protocol in terms of energy consumption and success rate. We compare our architecture to flooding, dynamic source routing (DSR), zone-routing protocol (ZRP), and two power-aware schemes. Our results show substantial power savings for our contact-based protocol, especially for large ad hoc networks. Index Terms--Adhoc networks, contact-based architecture, energy efficiency, network simulation, routing protocol.

A novel hybrid routing protocol - the Two-Zone Routing Protocol (TZRP) as a non-trivial extension of Zone Routing Protocol is proposed. TZRP uses two different zones to decouple the protocols ability to adapt to traffic characteristics from the ability to adapt to mobility.

A novel, efficient and low cost start-up and restart synchronization approach for TDMA environments is presented by focusing on bus-based communication protocols. The approach utilizes information about a node's message length that forms a unique sequence to achieve synchronization such that communication overhead can be avoided.

Different static and dynamic schemes are explored to select the virtual lanes for outgoing messages and mix the load among them without restricting any particular type of message to be carried only by a particular virtual network. In addition, the network latency, the number of message buffers per virtual lane, and the depth of network interface output queues affect the relative performance of various virtual lane selection algorithms are examined.

A new transmission protocol that uses a joint lossless-source and channel coding scheme (JARQ) is introduced. The protocol is evaluated by analyzing its throughput, and simulation results are presented that compare the throughput achieved with this protocol with that of conventional ARQ protocols.

The bulk synchronous task-scheduling problem (BSSPO) is known as an effective task scheduling problem for distributed-memory machines. A proof of NP-completeness of the decision counterpart of BSSPO is presented.

We present information theoretic results associated with digital cellular radio with special emphasis on time division multiple access (TDMA) protocols. In the considered model, the message is transmitted in K blocks (slots) each of duration T, where the signal occupies a bandwidth of W (Hz). The value of K is dictated by the decoding delay constraints, and we assume that 2WT [is much greater than] 1; that is, in each block, large block-length codewords (or large constraint length convolutional codes) are accommodated. In each block, it is assumed that the signal is received through a two (or a single, as a special case) ray propagation mechanism, where the scattered components are modeled by independent complex Gaussian processes of bandwidth B (Hz). It is also assumed that BT [is much lesser than] 1, and, therefore, the scattered components in each block are approximated by complex random variables which are in general correlated. The receiver is supposed to be fully aware of the channel parameters in all K blocks, while the transmitter does not possess this knowledge. For stringent decoding delay constraints (i.e., K is small) the system has no Shannon capacity in the strict sense, and a natural information-theoretic performance measure is based on the capacity versus outage probability characteristics, where the latter is determined by the probability that the instantaneous channel parameters cannot support the considered data rate. Average capacities are also considered, and it is shown that these are of little value when used to predict performance for small K and acceptably low outage probabilities. For K [approaches] [infinity] (absolutely relaxed delay constraints), however, and under mild conditions of decrease in correlation between the scattered gains in separated blocks, the average capacity yields the Shannon capacity in the usual sense. It is demonstrated that for small outage probabilities, which are of practical interest, the double ray propagation mechanism yields an advantage in the capacity versus outage performance sense when compared to a single ray propagation model, under the same signal-to-noise ratio ([Rho]) conditions. For K = 2, the performance of the optimal communication method is compared to that achieved by a simple time-diversity (two-fold repetition) with either selection or optimum ratio combining, and it is concluded that even these simple time-diversity methods are useful and enhance performance considerably. The advantage is performance for K = 2 over K = 1 is mitigated to a serious degree only for large values of the correlation ([Tau]) between the scattered components in the two blocks. Space diversity with two diversity branches is also addressed, reinterpreting the results developed for the two-fold time-diversity method.

The extent to which possible concurrent communication occurs in high speed wide area networks (WAN) is quantified by the performance metric, known as receptivity. Receptivity is the probability of all concurrently established connections for any stochastic demand pattern model. An exponential complexiety is exhibited by receptivity since the value of receptivity is estimated analytically.

We address the issue of using deformable models to reconstruct an unknown attenuation map of the torso from transmission scans. We focus on the reconstruction of a piecewise-uniform attenuation coefficients distribution. We represent the unknown distribution by a set of closed surfaces defining regions having the same attenuating properties. We show how free-form deformations allow one to deform a complex attenuation map template to match a set of transmission scans. The deformability of the model and the accurate estimation of the attenuating properties are closely studied in this paper.