Browsing by Author "Davis, Nathaniel J. IV"
Now showing 1 - 20 of 59
Results Per Page
Sort Options
- An access layer protocol for parallel networksKumar, Rajesh (Virginia Tech, 1993-05-15)Parallelism can be applied to local area networks to achieve higher data rates using existing hardware technologies. Parallelism can be employed at the different layers of the Open Systems Interconnection (OS1) reference model. This research proposes and analyzes a technique that permits the use of different degrees of parallelism at different protocol layers. A new protocol layer, called the access layer is defined. The protocol and functionality for this layer are defined. The definitions make provision for incorporating an error correction coding procedure known as cross channel coding. A software simulator was built for the proposed parallel network. The simulator has a detailed model of the access layer and was used to verify the functionality defined for the access layer and to estimate the performance of the parallel network. The simulation results indicate that although the access layer processes data in serial, it is not a bottleneck in the parallel system. Other insights obtained from the experiments are also presented.
- Active Library Resolution in Active NetworksLee, David C. (Virginia Tech, 1998-02-20)An active network is a computer network in which new protocols can be installed at run-time in any node within the network. For example, the deployment of Internet multicast technology has been slow because service providers have been reluctant to upgrade and reconfigure their routing nodes. Under the active network scheme, users who desire multicast services can have the service automatically installed without any direct intervention by the user or the provider. One major question in realizing active networks is how the code for the new active library can be found, or resolved, and retrieved. A model of the resolution and retrieval mechanisms is the major focus of this research. To validate the model, a proof-of-concept experimental system that realizes a simple active network architecture was developed. An active library resolution service model, suitable for a global Internet, was investigated using this experimental platform and a simulation system. The two protocol components that were built and studied are the active transport protocol and the active library resolution protocol. The experimental and simulation systems were used to evaluate the extensibility, overhead, resolution time, scalability, and policy constraint support of the service. Extensibility and policy constraint support are an integral part of the proposed design. For libraries located on servers that are at most ten hops away from the requesting source, the resolution time is under 2.6 seconds. Simulations of networks of different sizes and with different error rates exhibit linear resolution time and overhead characteristics, which indicates potential scalability. Behavior under high loss rates showed better than expected performance. The results indicate that the library resolution concept is feasible and that the proposed strategy is a good solution.
- Battery-Based Intrusion DetectionJacoby, Grant Arthur (Virginia Tech, 2005-04-12)This dissertation proposes an efficacious early warning system via a mobile host-based form of intrusion detection that can alert security administrators to protect their corporate network(s) by a novel technique that operates through the implementation of smart battery-based intrusion detection (B-bid) on mobile devices, such as PDAs, HandPCs and smart-phones by correlating attacks with their impact on device power consumption. A host intrusion detection engine (HIDE) monitors power behavior to detect potential intrusions by noting consumption irregularities and serves like a sensor to trigger other forms of protection. HIDE works in conjunction with a Scan Port Intrusion Engine (SPIE) that ascertains the IP and port source of the attack and with a host analysis signature trace engine (HASTE) that determines the energy signature of the attack and correlates it to a variety of the most common attacks to provide additional protection and alerts to both mobile hosts and their network.
- Cater: An Opportunistic Medium Access Control Protocol for Wireless Local Area NetworksMullins, Barry E. (Virginia Tech, 1997-06-23)An adaptive MAC protocol is developed and analyzed that offers a "best case" scenario by allowing the MAC to control medium parameters thereby fully exploiting the channel of an ad hoc wireless LAN. This new, opportunistic medium access control protocol is called CATER (Code Adapts To Enhance Reliability) and is based on the proposed MAC standard for wireless local area networks (WLAN)-IEEE 802.11 [IEE96]. As currently proposed, IEEE 802.11 uses a fixed pseudo-noise (PN) code for spreading the information signal, implying a fixed process gain at the receiver. When the channel degrades, IEEE 802.11 offers only retransmissions at the MAC layer to combat a corrupt medium. However, CATER allows communicating stations to reconfigure their transceivers to use a longer PN code after a prescribed number of failed retransmissions. This longer code increases the process gain of the receiver and reduces the error rate. After the two stations are reconfigured, the source station sends the frame in question. Immediately after that frame is acknowledged, the source station may send additional frames during the reconfigured period. Simulation and emulation are used to demonstrate and validate the adaptive protocol's capabilities. Results show that this new protocol offers substantial improvement in system throughput when the channel degrades to a point that reliable transmission of frames is not feasible in a standard IEEE 802.11 WLAN. Specifically, CATER continues to function, permitting up to 14 percent normalized aggregate throughput at times when IEEE 802.11 permits no frames to pass through the WLAN. In addition, throughput experiences only a small decrease due to protocol overhead during periods when stations experience a good channel with few bit errors. Moreover, CATER does not adversely affect the predominate transport layer protocol (i.e., TCP), and provides equitable service to all stations within the network.
- Context Aware and Adaptive Security for Wireless NetworksHager, Creighton Tsuan-Ren (Virginia Tech, 2004-11-30)This research investigated methods to determine appropriate security protocols for specific wireless network applications. The specific problem being addressed was that there are tradeoffs between security, performance, and efficiency among current and proposed security protocols. Performance and efficiency issues are particularly important in wireless networks which tend to have constrained network capacity and connect to resource-limited nodes. Existing security protocols address problems such as authentication, availability, confidentiality, integrity, and non-repudiation. However, these protocols use resources and limit the efficient use of node resources. Thus, the overall objective of this research is to improve the efficiency of security mechanisms for wireless networks. A methodology was constructed to satisfy this objective and is an important contribution of this research. The methodology can be used to define the relevant operational parameters of different wireless network applications, classify wireless networks into distinct categories, incorporate appropriate security protocols to a category, and analyze the security protocols through metrics. Three groups of operational parameters were created to classify wireless networks; these are equipment, network topology, and communication characteristics. The wireless network categories include, but are not limited to, fixed broadband wireless networks, wireless local area networks, mobile ad hoc networks, and small device sensor networks. The metrics in the methodology are used to measure end-to-end data throughput and delay, efficiency and overhead, power and energy consumption, and energy consumed per packet transferred. The main advantage of this methodology is the flexibility of how constraints are considered and suitability is analyzed. This approach can identify problems from manageable categories of networks and find or create solutions for each of them. Another advantage of this methodology is that after suitable security protocols are found or created for each category, any new wireless network application that falls into an existing category may be able to use the security protocols from that category and find that they are the most suitable. Another key contribution of this research was the implementation and evaluation of a context aware and adaptive security manager (CASM) that selects appropriate protocols in real-time. CASM was developed using the methodology as a guide. Results from a resource analysis of four encryption algorithms were utilized for the design of CASM. A feasibility study of CASM was then completed. Three different experimental scenarios were used to evaluate CASM's operation. The results and analysis of the experiments indicate that the security manager functions properly and security is provided efficiently with different user settings and environments. Three schemes were deemed the best to use for the decision module of CASM.
- Dataflow Analysis and Optimization of High Level Language Code for Hardware-Software Co-DesignO'Connor, R. Brendan (Virginia Tech, 1996-05-07)Recent advancements in FPGA technology have provided devices which are not only suited for digital logic prototyping, but also are capable of implementing complex computations. The use of these devices in multi-FPGA Custom Computing Machines (CCMs) has provided the potential to execute large sections of programs entirely in custom hardware which can provide a substantial speedup over execution in a general-purpose sequential processor. Unfortunately, the development tools currently available for CCMs do not allow users to easily configure multi-FPGA platforms. In order to exploit the capabilities of such an architecture, a procedure has been developed to perform a dataflow analysis of programs written in C which is capable of several hardware-specific optimizations. This, together with other software tools developed for this purpose, allows CCMs and their host processors to be targeted from the same high-level specification.
- Decentralized Trust-Based Access Control for Dynamic Collaborative EnvironmentsAdams, William Joseph (Virginia Tech, 2006-03-31)The goal of this research was to create a decentralized trust-based access control (TBAC) system for a dynamic collaborative environment (DCE). By building a privilege management infrastructure (PMI) based on trust, user access was determined using behavior grading without the need for pre-configured, centrally managed role hierarchies or permission sets. The PMI provided TBAC suitable for deployment in a rapidly assembled, highly fluid, collaborative environment. DCEs were assembled and changed membership as required to achieve the goals of the group. A feature of these environments was that there was no way of knowing who would join the group, no way of refusing anyone entry into group, and no way of determining how long members would remain in the group. DCEs were formed quickly to enable participants to share information while, at the same time, allowing them to retain control over the resources that they brought with them to the coalition. This research progressed the state of the art in the fields of access control and trust management. The Trust Management System developed through this research effectively implemented a decentralized access control scheme. Each resource owner independently evaluated the reputation and risk of network members to make access decisions. Because the PMI system used past behavior as an indication of future performance, no a priori user or resource configuration was required.
- Decoder Board Hardware/Software Development in Wireless Interactive Video Data Service SystemGoedde, Todd William (Virginia Tech, 1997-09-18)The Interactive Video Data Service (IVDS) system allows consumers to browse the Internet, request information on products or services, make purchases, indicate preferences, and perform other interactive applications. To provide this service, the IVDS system has three subsystems: Consumer Control (CC), Cell Repeater (CR), and Host subsystem. In the CC subsystem, an IVDS transceiver box is placed near a television set. Once the consumer sends a command to the transceiver box using a standard television/VCR/Cable remote control, the transceiver box receives information embedded in the television audio, and then transmits the information to the CR subsystem as a radio frequency (RF) spread spectrum message. The CR subsystem decodes the spread spectrum message and forwards it to the Host subsystem for processing. Located in the CR subsystem, a custom designed circuit board, called the decoder board, uses surface mounted components to decode and packetize the spread spectrum message for transfer to the CR main processor. This paper provides a functional description of the hardware components on the decoder board, and describes the hardware/software developed for interfacing the decoder board to the radio receiver and to the CR main processor. Hardware modifications were needed to correct timing problems between components. Software was developed to initialize the components for downconverting, despreading, and demodulating spread spectrum messages, and to packetize them for transfer to the CR main processor. This paper also discusses the tests used to verify both the performance of the decoder board software and the operation of the hardware components.
- The Design and Implementation of a Nanosatellite State-of-Health Monitoring SubsystemBolton, Bryce Daniel (Virginia Tech, 2001-12-06)This research consists of the design of a low-power, low-cost, nanosatellite computer system solution. The proposed system solution, and design and implementation of a multiple-bus master FPGA and health monitoring space computer subsystem are described. In the fall of 1998, the US Air Force (USAF) funded Virginia Polytechnic Institute & State University (Virginia Tech), The University of Washington (UW), and Utah State University (USU) with $100,000 each to pursue a formation-flying satellite cluster. The program specified that a cluster of three satellites would maintain radio contact through UHF cross-link communication to report relative positions, obtained through GPS, and coordinate scientific measurement mission activities. This satellite cluster, named Ionospheric Observation Nanosatellite Formation (ION-F) is presently scheduled for launch in June of 2003. Maintaining some degree of system reliability in the error-prone space environment was desired for this low-cost space program. By utilizing high-reliability components in key system locations, and monitoring less reliable portions of the computer system for faults, an improvement in overall system reliability was achieved. The development of a one-wire health monitoring bus master was performed. A Synchronous Serial Peripheral Interface (SPI) bus master was utilized to extend the communication capabilities of the CPU. In addition, discrete I/O functions and A/D converter interfaces were developed for system health monitoring and the spacecraft Attitude Determination and Control System (ADCS).
- Design evaluation of alternative sonar system configurationsBalash, Daniel J. (Virginia Tech, 1995-07-01)The United States Navy has concentrated the majority of its Surface Anti Submarine Warfare (ASW) capability in systems that are designed to exploit the specific acoustic environment of the open ocean. The new priorities are the littoral threat of mined harbors, gulfs, bays, and coastal areas and of coastal patrolling diesel submarines operating in a familiar, although more complex environment. The Naval operational requirements are to exercise control of threat platforms at ranges greater than the threat's weapon range capability and to provide defense against mobile (torpedoes) and immobile (mines) weapons. The United States' continued intervention in small skirmishes around the world requires an active sonar system that can perform the roles of shallow water and small object avoidance. An engineering approach to develop the new sonar system that can meet the defined operational requirements is needed. The system will be the first mission critical combat system built from a majority of COTS components in the USN inventory and will provide small object avoidance and shallow water capabilities to combatants where none previously existed.
- Design of a hardware interface for a high-speed parallel networkHarper, Scott Jeffery (Virginia Tech, 1994-08-15)Parallelism can use existing technology in computer communications network design to provide higher data rates and a greater degree of flexibility than monolithic systems. This research investigates the design of a high-speed Parallel Local Area Network (PLAN) interface. It defines the goals of a PLAN interface as low data latency, high data throughput, scalability, and low cost. Three fundamental PLAN interface categories are proposed to meet these goals. These categories are single-bus, dual-bus, and bus-free adaptors. The relative merits of each category are discussed in terms of suitability to several adaptor applications. Each category is further explored by developing a VHDL model of a representative system. The latency, throughput, and component utilization of each model is measured. For medium to large data sets, the dual-bus design provides slightly greater throughput when transmitting encoded data. When transmitting medium to large unencoded data sets, the bus-free design yields marginally higher throughput. In nearly all cases the bus-free design has a greater latency than either of the bus-based design options. Other insights gained from the models regarding physical construction of each adaptor type are also presented.
- The Design of an IVDS World Wide Web Browser ArchitectureHawes, Aaron George (Virginia Tech, 1997-11-20)An IVDS (Interactive Video Data Service) uses an interactive television system to transmit data to and from subscribers' homes. IVDS allows the viewer to interact with content provided on the television using a remote control. A typical IVDS application would be ordering an advertised product or playing along with a quiz show. The Virginia Tech Center for Wireless Telecommunications (CWT), under a contract with Interactive Return Service, Inc., is developing an IVDS system in which content is provided through the television cable system in the form of audio codes. A special remote control can detected these audio codes and query the user for input. The return path for this system is a wireless channel. The remote control contains a spread spectrum transmitter that transmits packets to a Repeater unit residing within a quarter mile of the user's home. With the popularity of the World Wide Web soaring, many companies are announcing internet appliances that will bring the content of the web to the user at a fraction of the cost of a standard personal computer. CWT has been contracted to extend the core IVDS system to provide a web browsing capability, allowing the user to browse the web with only the remote control. This thesis outlines the requirements of the IVDS Web Browser System. The different hardware design concepts are documented. The final Browser System specification is presented, as well as a board-level description of the Decoder Unit that is part of this final Browser System. Finally, a detailed description, current status, and simulation results are presented for the FPGA (Field Programmable Gate Array) that serves as the controller for the Decoder Unit.
- Designing a Compiler for a Distributed Memory Parallel Computing SystemBennett, Sidney Page (Virginia Tech, 2003-11-01)The SCMP processor presents a unique approach to processor design: integrating multiple processors, a network, and memory onto a single chip. The benefits to this design include a reduction in overhead incurred by synchronization, communication, and memory accesses. To properly determine its effectiveness, the SCMP architecture must be exercised under a wide variety of workloads, creating the need for a variety of applications. A compiler can relieve the time spent developing these applications by allowing the use of languages such as C and Fortran. However, compiler development is a research area in its own right, requiring extensive knowledge of the architecture to make good use of its resources. This thesis presents the design and implementation of a compiler for the SCMP architecture. The thesis includes an in-depth analysis of SCMP and the necessary design choices for an effective compiler using the SUIF and MachSUIF toolsets. Two optimizations passes are included in the discussion: partial redundancy elimination and instruction scheduling. While these optimizations are not specific to parallel computing, architectural considerations must still be made to properly implement the algorithms within the SCMP compiler. These optimizations yield an overall reduction in execution time of 15-36%.
- Detecting software attacks by monitoring electric power consumption patterns(United States Patent and Trademark Office, 2011-01-25)Software attacks such as worms and viruses are detected in an electronic device by monitoring power consumption patterns. In a first embodiment, software attacks are detected by an increase in power consumption. The increased power consumption can be caused by increased network traffic, or by increased activity in the microprocessor. Monitoring power consumption is particularly effective for detecting DOS/flooding attacks when the electronic device is in an idle state. In a second embodiment, a power consumption signal is converted to the frequency domain (e.g., by fast Fourier transform). The highest amplitude frequencies are identified. Specific software attacks produce characteristic frequencies in the power consumption signal. Software attacks are therefore detected by matching the highest amplitude frequencies with frequencies associated with specific worms and viruses. Identification of a particular software attack typically requires matching of 3 or more of the highest amplitude frequencies, and, optionally, amplitude information.
- Dynamic Code Sharing Algorithms for IP Quality of Service in Wideband CDMA 3G Wireless NetworksFossa, Carl Edward Jr. (Virginia Tech, 2002-04-22)This research investigated the efficient utilization of wireless bandwidth in Code Division Multiple Access (CDMA)systems that support multiple data rates with Orthogonal Variable Spreading Factor (OVSF)codes. The specific problem being addressed was that currently proposed public-domain algorithms for assigning OVSF codes make inefficient use of wireless bandwidth for bursty data traffic sources with different Quality of Service (QoS) requirements. The purpose of this research was to develop an algorithm for the assignment of OVSF spreading codes in a Third-Generation (3G)Wideband CDMA (WCDMA)system. The goal of this algorithm was to efficiently utilize limited, wireless resources for bursty data traffic sources with different QoS requirements. The key contribution of this research was the implementation and testing of two code sharing techniques which are not implemented in existing OVSF code assignment algorithms. These techniques were termed statistical multiplexing and dynamic code sharing. The statistical multiplexing technique used a shared channel to support multiple bursty traffic sources. The dynamic code sharing technique supported multiple data users by temporarily granting access to dedicated channels. These techniques differed in terms of both complexity and performance guarantees.
- Energy-efficient Wireless Sensor Network MAC ProtocolBrownfield, Michael I. (Virginia Tech, 2006-03-31)With the progression of computer networks extending boundaries and joining distant locations, wireless sensor networks (WSNs) emerge as the new frontier in developing opportunities to collect and process data from remote locations. WSNs rely on hardware simplicity to make sensor field deployments both affordable and long-lasting without maintenance support. WSN designers strive to extend network lifetimes while meeting application-specific throughput and latency requirements. Effective power management places sensor nodes into one of the available energy-saving modes based upon the sleep period duration and the current state of the radio. This research investigates energy-efficient medium access control (MAC) protocols designed to extend both the lifetime and range of wireless sensor networks. These networks are deployed in remote locations with limited processor capabilities, memory capacities, and battery supplies. The purpose of this research is to develop a new medium access control protocol which performs both cluster management and inter-network gateway functions in an energy-efficient manner. This new protocol, Gateway MAC (GMAC), improves on existing sensor MAC protocols by not only creating additional opportunities to place the sensor platforms into lower power-saving modes, but also by establishing a traffic rhythm which extends the sleep duration to minimize power mode transition costs. Additionally, this research develops a radio power management (RPM) algorithm to provide a new mechanism for all WSN MAC protocols to optimize sleep transition decisions based upon the power and response characteristics of the sensor platform's transceiver. Finally, to extend access to sensor data in remote locations, this research also validates an innovative wireless distribution system which integrates wireless sensor networks, mobile ad hoc networks (MANET), and the Internet. This research makes two significant contributions to the state-of-the-art wireless sensor networks. First, GMAC's centralized network management function offers significant energy savings and network lifetime extensions over existing wireless sensor network protocols. The second contribution is the introduction of a wireless sensor radio power management algorithm designed to exploit additional power-saving opportunities introduced with the newest generation of faster sensor platform transceivers.
- Extensions for Multicast in Mobile Ad-hoc Networks (XMMAN): The Reduction of Data Overhead in Wireless Multicast TreesChristman, Michael Edward (Virginia Tech, 2002-07-31)Mobile Ad hoc Network (MANET) routing protocols are designed to provide connectivity between wireless mobile nodes that do not have access to high-speed backbone networks. While many unicast MANET protocols have been explored, research involving multicast protocols has been limited. Existing multicast algorithms attempt to reduce routing overhead, but few, if any, attempt to reduce data overhead. The broadcast nature of wireless communication creates a unique environment in which overlaps in coverage are common. When designed properly, a multicast algorithm can take advantage of these overlaps and reduce data overhead. Unlike a unicast route, in which there is one path between a sender and receiver, a multicast tree can have multiple branches between the sender and its multiple receivers. Some of these paths can be combined to reduce redundant data rebroadcasts. The extensions presented in this thesis are a combination of existing and original routing techniques that were designed to reduce data rebroadcasts by aggregating multicast data flows. One such optimization takes advantage of the multipoint relay (MPR) nodes used by the Optimized Link State Routing (OLSR) unicast protocol. These nodes are used in unicast routing to reduce network broadcast, but can be used to help create efficient multicast data flows. Additionally, by listening to routing messages meant for other nodes, a host can learn a bit about its network and may be able to make routing changes that improve the multicast tree. This protocol was implemented as a software router in Linux. It should be emphasized that this is a real implementation and not a simulation. Experiments showed that the number of data packets in the network could be reduced by as much as 19 percent. These improvements were accomplished while using only a small amount of routing overhead.
- The Folded Hypercube ATM SwitchesPark, Jahng Sun (Virginia Tech, 2001-09-18)Over the past few years, many high performance asynchronous transfer mode (ATM) switches have been proposed. The majority of these switches have high performance but also high hardware complexity. Therefore, there is a need for switch designs with low complexity and high performance. This research proposes three new ATM switches based on the folded hypercube network (FHC). The performance of the three architectures are studied using a network model and simulation. The major performance parameters measured are the cell loss rate and cell delay time through the switch under uniform, normal, and bursty traffic patterns. To guarantee faster switching of time-sensitive cells, the routing algorithm of the three switches uses a priority scheme that gives higher precedence to the time-sensitive cells. Also, an output buffer controller is designed to manage the buffers in a fair manner. The three proposed switch architectures have lower complexity while providing equivalent or better switching performance compared to other more complex ATM switches described in the literature. This research shows a new approach to designing ATM switches by using the FHC as the switching fabric for the first time instead of using the crossbar, multi-path, or Banyan-based switching fabrics.
- A Framework for Providing Redundancy and Robustness in Key Management for IPsec Security Associations in a Mobile Ad-Hoc EnvironmentHadjichristofi, George Costa (Virginia Tech, 2005-06-28)This research investigated key management in a Mobile Ad Hoc Network (MANET) environment. At the time this research began key management schemes provided limited functionality and low service availability in a highly partitioned ad hoc environment. The purpose of this research was to develop a framework that provides redundancy and robustness for Security Association (SA) establishment between pairs of nodes. The key contribution of this research is the Key Management System (KMS) framework and, more specifically, the unique way the various components are integrated to provide the various functionalities. The KMS overcomes the limitations of previous systems by (1) minimizing pre-configuration, (2) increasing service availability, (3) and increasing flexibility for new nodes joining the network. A behavior grading scheme provides the network with a system-wide view of the trustworthiness of nodes and enables the KMS to dynamically adjust its configuration according to its environment. The introduction of behavior grading allows nodes to be less dependent on strict identity verification. This KMS was simulated with Monte Carlo and NS2 simulations and was shown to interoperate with IP Security (IPsec) to enable the establishment of IPsec SAs. The simulations have proven the effectiveness of the system in providing service to the nodes in a highly partitioned environment.
- Fundamentals of the Simplex Communication Channel With RetransmissionsDavidson, Boris (Virginia Tech, 1997-04-14)The need for multiple access strategies arises whenever a number of users have to share a communication resource, since it is usually either cost prohibitive or impractical to dedicate a communication channel to a particular user. A need for such algorithms arises in many instances, particularly in applications utilizing wireless systems where all users access a common channel or medium. Such random access techniques as ALOHA and slotted ALOHA have been successfully implemented in a number of wireless applications. One of the major drawbacks of these algorithms is the necessity of a return path from the central station to each system user, which makes their use both inefficient and expensive for applications where one-way communication would suffice. For such applications, a need remained for a random access algorithm which can maximize the probability of successful message transmission in a one-way communication environment. A random access technique that addresses the above-mentioned need is developed. With this technique, each user sends an original message of predetermined length to a central receiver. The user then retransmits the message a specified number of times in a predetermined interval reserved for the retransmission process. The time interval between each successive retransmission of a given message is random. Assuming total annihilation of all colliding messages, the expression for the probability of successful transmission of a given message in terms of the major channel parameters is theoretically formulated. This technique offers a significant improvement, compared to a single transmission, in ensuring that a message is successfully received. The actual message collision dynamics in this system are experimentally studied using two different types of direct-sequence spread spectrum receivers, one employing a sliding correlator and the other using a matched filter. The spreading code in such systems offers extra protection for messages against possible interferers. The results indicate that it is often possible to properly receive a given message in the presence of co-channel interferers, thus significantly improving the overall system performance. These results are subsequently incorporated with the propagation data for several different types of microcells to arrive at a more precise theory of the link.
- «
- 1 (current)
- 2
- 3
- »