The most exciting development in parallel computer architecture is the convergence of traditionally disparate approaches on a common machine structure. This book explains the forces behind this convergence of shared-memory, message-passing, data parallel, and data-driven computing architectures. It then examines the design issues that are critical to all parallel architecture across the full range of modern design, covering data access, communication performance, coordination of cooperative work, and correct implementation of useful semantics. It not only describes the hardware and software techniques for addressing each of these issues but also explores how these techniques interact in the same system. Examining architecture from an application-driven perspective, it provides comprehensive discussions of parallel programming for high performance and of workload-driven evaluation, based on understanding hardware-software interactions.


* synthesizes a decade of research and development for practicing engineers, graduate students, and researchers in parallel computer architecture, system software, and applications development

* presents in-depth application case studies from computer graphics, computational science and engineering, and data mining to demonstrate sound quantitative evaluation of design trade-offs 

* describes the process of programming for performance, including both the architecture-independent and architecture-dependent aspects, with examples and case-studies

* illustrates bus-based and network-based parallel systems with case studies of more than a dozen important commercial designs

Table of Contents

1 Introduction 
2 Parallel Programs 
3 Programming for Performance 
4 Workload-Driven Evaluation 
5 Shared Memory Multiprocessors 
6 Snoop-based Multiprocessor Design 
7 Scalable Multiprocessors
8 Directory-based Cache Coherence
9 Hardware-Software Tradeoffs 
10 Interconnection Network Design 
11 Latency Tolerance 
12 Future Directions 
APPENDIX A Parallel Benchmark Suites

Parallel Computer Architecture: A Hardware/Software Approach

The major challenge in designing wireless sensor networks (WSNs) is the support of the functional, such as data latency, and the non-functional, such as data integrity, requirements while coping with the computation, energy and communication constraints. Careful node placement can be a very effective optimization means for achieving the desired design goals. In this paper, we report on the current state of the research on optimized node placement in WSNs. We highlight the issues, identify the various objectives and enumerate the different models and formulations. We categorize the placement strategies into static and dynamic depending on whether the optimization is performed at the time of deployment or while the network is operational, respectively. We further classify the published techniques based on the role that the node plays in the network and the primary performance objective considered. The paper also highlights open problems in this area of research.

/pdf/strategies-and-techniques-for-node-placement-in-wireless-49h7ns6zrt.pdf

Strategies and techniques for node placement in wireless sensor networks: A survey

Wireless sensors and wireless sensor networks have come to the forefront of the scientific community recently. This is the consequence of engineering increasingly smaller sized devices, which enable many applications. The use of these sensors and the possibility of organizing them into networks have revealed many research issues and have highlighted new ways to cope with certain problems. In this paper, different applications areas where the use of such sensor networks has been proposed are surveyed

/pdf/a-survey-of-applications-of-wireless-sensors-and-wireless-2uaprfs4xy.pdf

A Survey of Applications of Wireless Sensors and Wireless Sensor Networks

An Internet access device (100) uses an automatic configuration process (600) to handle the task of configuring the Internet access device at a customer site for communication with the Internet (10). Once configured, the customer has electronic mail and other access to the Internet from his local area network. A not yet configured Internet access device is shipped directly to a customer without having to be manually configured first. The customer enters a registration identification number (326) and a telephone number onto the Internet access device. The Internet access device then automatically connects to the Internet, downloads configuration data from a configuration server (410) containing customer site specific configuration data, and then automatically configures itself for communication with the Internet. The Internet access device is simple to install for a customer and provides valuable features such as a router (240), firewall, e-mail gateway (212), web server (220), and other servers (222). The Internet access device initially connects to the Internet through an Internet service provider (14) over a standard analog telephone line using a standard modem (52) and using a dynamic IP address. Once automatically configured, the Internet access device may then communicate with the Internet using any suitable connection including an analog telephone line, or a higher-speed line such as an ISDN line or a frame relay circuit and is assigned a static IP address and a range of IP addresses for other devices on its local area network.

Automatic configuration for internet access device

Appearing in 1st IEEE International Workshop on Sensor Net Protocols and Applications (SNPA). Anchorage, Alaska, USA. May 11, 2003. RMST: Reliable Data Transport in Sensor Networks Fred Stann, John Heidemann Abstract – Reliable data transport in wireless sensor networks is a multifaceted problem influenced by the physical, MAC, network, and transport layers. Because sensor networks are subject to strict resource constraints and are deployed by single organizations, they encourage revisiting traditional layering and are less bound by standardized placement of services such as reliability. This paper presents analysis and experiments resulting in specific recommendations for implementing reliable data transport in sensor nets. To explore reliability at the transport layer, we present RMST (Reliable Multi- Segment Transport), a new transport layer for Directed Diffusion. RMST provides guaranteed delivery and fragmentation/reassembly for applications that require them. RMST is a selective NACK-based protocol that can be configured for in-network caching and repair. Second, these energy constraints, plus relatively low wireless bandwidths, make in-network processing both feasible and desirable [3]. Third, because nodes in sensor networks are usually collaborating towards a common task, rather than representing independent users, optimization of the shared network focuses on throughput rather than fairness. Finally, because sensor networks are often deployed by a single organization with inexpensive hardware, there is less need for interoperability with existing standards. For all of these reasons, sensor networks provide an environment that encourages rethinking the structure of traditional communications protocols. The main contribution is an evaluation of the placement of reliability for data transport at different levels of the protocol stack. We consider implementing reliability in the MAC, transport layer, application, and combinations of these. We conclude that reliability is important at the MAC layer and the transport layer. MAC-level reliability is important not just to provide hop-by-hop error recovery for the transport layer, but also because it is needed for route discovery and maintenance. (This conclusion differs from previous studies in reliability for sensor nets that did not simulate routing. [4]) Second, we have developed RMST (Reliable Multi-Segment Transport), a new transport layer, in order to understand the role of in- network processing for reliable data transfer. RMST benefits from diffusion routing, adding minimal additional control traffic. RMST guarantees delivery, even when multiple hops exhibit very high error rates. 1 Introduction Wireless sensor networks provide an economical, fully distributed, sensing and computing solution for environments where conventional networks are impractical. This paper explores the design decisions related to providing reliable data transport in sensor nets. The reliable data transport problem in sensor nets is multi-faceted. The emphasis on energy conservation in sensor nets implies that poor paths should not be artificially bolstered via mechanisms such as MAC layer ARQ during route discovery and path selection [1]. Path maintenance, on the other hand, benefits from well- engineered recovery either at the MAC layer or the transport layer, or both. Recovery should not be costly however, since many applications in sensor nets are impervious to occasional packet loss, relying on the regular delivery of coarse-grained event descriptions. Other applications require loss detection and repair. These aspects of reliable data transport include the provision of guaranteed delivery and fragmentation/ reassembly of data entities larger than the network MTU. Sensor networks have different constraints than traditional wired nets. First, energy constraints are paramount in sensor networks since nodes can often not be recharged, so any wasted energy shortens their useful lifetime [2]. This work was supported by DARPA under grant DABT63-99-1-0011 as part of the SCAADS project, and was also made possible in part due to support from Intel Corporation and Xerox Corporation. Fred Stann and John Heidemann are with USC/Information Sciences Institute, 4676 Admiralty Way, Marina Del Rey, CA, USA E-mail: fstann@usc.edu, johnh@isi.edu. 2 Architectural Choices There are a number of key areas to consider when engineering reliability for sensor nets. Many current sensor networks exhibit high loss rates compared to wired networks (2% to 30% to immediate neighbors)[1,5,6]. While error detection and correction at the physical layer are important, approaches at the MAC layer and higher adapt well to the very wide range of loss rates seen in sensor networks and are the focus of this paper. MAC layer protocols can ameliorate PHY layer unreliability, and transport layers can guarantee delivery. An important question for this paper is the trade off between implementation of reliability at the MAC layer (i.e. hop to hop) vs. the Transport layer, which has traditionally been concerned with end-to-end reliability. Because sensor net applications are distributed, we also considered implementing reliability at the application layer. Our goal is to minimize the cost of repair in terms of transmission.

RMST: Reliable Data Transport in Sensor Networks

We describe a wireless sensor network designed for the long-term study of rare and endangered species of plants We wish to monitor plants and their environment via high-resolution cameras and temperature, humidity, rainfall, wind, and solar radiation sensors Our units must be A¢Â€ÂœinvisibleA¢Â€Â camouflaged, very low energy, and must allow distributed local computation Data rates are 1 to 100 bytes/second per node, but networks can be large - an early prototype had 60 nodes Failures are expensive and we must exploit redundancy whenever possible Nodes are stationary but for energy reasons may decline to participate in transmissions

We have designed two wireless routing protocols that satisfy these constraints Multipath On-demand Routing MOR computes multiple optimal routes to avoid depleting the energy at any given node Geometric Routing scales to large networks, relying on Geographic Routing when possible and on selected global information otherwise We have simulated and are implementing both protocols

The Application of Remote Sensor Technology To Assist the Recovery of Rare and Endangered Species

Sensors can be paired with radio units and deployed to form a wireless ad-hoc sensor network. Actual deployments must consider the coverage that can be achieved with a given number of sensors: this coverage varies with the range of the radios and the maximum allowable distance between any point in the area and the nearest sensor. Deployments must also preserve connectivity in spite of possible failure or energy depletion in a subset of the units. This paper presents and analyzes a variety of regular deployment topologies, including circular and star deployments as well as deployments in square, triangular, and hexagonal grids.

/pdf/wireless-sensor-placement-for-reliable-and-efficient-data-53dn1ywwca.pdf

Wireless sensor placement for reliable and efficient data collection

The adaptor cards and driver software for workstations and local asynchronous transfer mode (ATM) switches and switch control software used in an ATM local area network (LAN) system are discussed. It is shown that the ATM hardware and software components together provide services that are essential for ATM to be considered a realistic alternative to current shared-media LANs. These services include: completely transparent support for the TCP/IP protocol suite, an application programming interface for full access to the underlying ATM capabilities, support for AAL5, AAL3/4, and the null AAL, both connection-oriented and connectionless service, dynamic connection establishment or switched virtual circuits, resource reservation of guaranteed bandwidth and quality of service, full-bandwidth multicast and broadcast, virtual path and channel routing among multiple switches, automatic configuration and failure recovery, dynamic address assignment and internetwork address resolution, and network management via the simple network management protocol (SNMP). >

Designing a practical ATM LAN

The articles in this special section focus on ad hoc and sensor networks. Today we are witnessing an interesting paradigm shift from the traditional client-server computing model, and the Internet of Things (IoT) is driving it. The reality is that the cost of Internet connectivity for wearable and mobile technology is decreasing, and our daily routines depend more on the use of devices that come equipped with more sensors and much more interesting capabilities at much lower costs. As a result, sensors, connected machinery, and many other things are all becoming connected devices. New enterprises take opportunity from this by exploiting data being collected from such connected devices, for new services and products designed to react to data in the most personalized ways. There is a growing trend in companies dominating their respective industries without owning estates or tangible assets, all thanks to data.

Ad hoc and sensor networks

Advanced programming languages such as Standard ML have rarely been used for systems programming tasks such as operating systems and network communications. In order to understand more fully the requirements of systems programming, we have implemented a suite of industry-standard network communication protocols in a completely type-safe extension of Standard ML. While the implementation has only recently become operational, we already observe acceptable communications throughput. We make careful use of the Standard ML modules system, with the core component of the implementation being a signature which is generic to all communications protocols. This generic protocol is then specialized for specific protocols, and these are implemented by functors parameterized by generic protocols. This leads naturally to a layered system structure and also provides an important and useful “mix-and-match” capability in composing protocols into complex networking systems.We have found the advanced features of Standard ML, in particular the modules system, static typing, and higher-order functions, to be extremely useful in building complex communications systems. The type compatibility of the various components of a system is guaranteed by the compiler. Furthermore, we find it significant that most of the information needed to understand the structure and interactions in our code can be obtained from a study of the signatures alone. Perhaps most important is that we have been able to use the expressive power of Standard ML modules to give concrete expression to previously ad hoc system-structuring concepts developed by other researchers in the field of network communications. For language designers and implementors, our experience has also pointed out specific areas for further work that may lead to advanced languages that are useful for systems programming.

Edoardo Biagioni

Papers

The Application of Remote Sensor Technology To Assist the Recovery of Rare and Endangered Species

Wireless sensor placement for reliable and efficient data collection

Designing a practical ATM LAN

Ad hoc and sensor networks

Signatures for a network protocol stack: a systems application of Standard ML