Keith Olds

Quantifying the relative merits of genetic and swarm algorithms for network optimization in cognitive radio networks

Cognitive engines have been under study and development for a number of years as a technique for addressing the needs of cognitive radios [1,2,3]. More recently there has been effort to expand the role of the cognitive engine to address the needs of a network of cognitive radios [4,5]. Haykin [6] has demonstrated that there is a significant difference between a network of cognitive radios and a cognitive radio network. This paper addresses three questions: 1. What are the significant functional and parametric differences between cognitive algorithms that deal with optimizing the operations of a cognitive radio and cognitive algorithms that optimize the operations of a cognitive radio network? 2. What are the trade-offs in applying the various algorithms to each task? 3. Which algorithms are optimal for the networking tasks? This paper identifies a set of parameters that characterize candidate algorithms and explores the benefits and drawbacks of each for cognitive network tasks. We propose a tiered architecture of cognitive engine algorithms that work in tandem to optimize the use of cognitive networked radios for the optimal success of the networked mission.

The DirecNet™ standard reference architecture: A roadmap for interoperability

The Open Groups DirecNet™ Task Force is an industry-led consortium consisting of twelve major aerospace firms. Its goal is to develop a vendor-neutral interoperability standard for a next generation directional networking waveform. The waveform developed will be IP-enabled, support the use of directional high bandwidth links, and the use of ad hoc mobile mesh networking. A key objective of the Task Force is to bring the advantages of commercial interoperability and standards development processes into the government arena. At this time, the Task Force has completed and released an overview and reference architecture volume as well the standard for a bandwidth-efficient PHY layer. This paper reviews the selected DirecNet architecture and on-going technical work of the consortium.

A comparison of FDD and TDD/TDMA architectures for airborne backbone network traffic

An important aspect of extending the DOD GIG into the tactical battle-space will be the development of high capacity airborne backbone IP networks. Properly designing and provisioning these networks is critically dependent on the statistical characteristics of the data flows that these networks will encounter. This paper studies the architecture of general backbone networks and airborne backbone networks through discrete event simulation methods. A network composed of FDD (frequency division duplex) links is quantitatively compared to a network composed of TDD (time division duplex) links using several criteria. Performance models for both architectures when carrying conventional and self-similar traffic were developed. The effectiveness of provisioning with open loop control was quantified as well as dynamic bandwidth allocation with closed loop control.

Multi-Function Highband Software Radio with Digital Transceivers

In this paper we describe modular hardware architecture for a mobile ad hoc networking software defined radio that is waveform protocol agnostic. The present plan for the current hardware is to host at least two different waveforms - terrestrial/airborne and SATCOM, but the architecture is extensible for many more waveforms. The waveforms are hosted on a Highband Digital Transceiver (HDT), which provides three processing engines. Two of the engines are Modified-Commercial-Off-The-Shelf (MOTS) compute engines. The remaining engine is a three field programmable gate array (FPGA) signal processor, one general purpose processor (GPP), 1 Gbps Ethernet input, 10 Gbps Ethernet output, reduced latency resident memory, fiber optic connected modular platform. All three engines are housed in a custom enclosure that inserts into a JTRS-like form factor vehicle adaptor. The HDT is a programmable device that can be configured through appropriate software loading to support the core waveforms such as the High-band Networking Waveform (HNW - terrestrial/airborne) and the Network Centric Waveform (NCW - SATCOM). An HDTs purpose is to provide part of the physical layer (PHY), media access control layer (MAC), and network layer (NET) processing for the terrestrial/airborne and SATCOM waveforms.

Real-time control and management of roaming wideband SATCOM terminals

Wideband SATCOM provides essential beyond line-of-sight communications services. At this time, however, no widely adopted protocols exist to facilitate roaming between access domains. In addition, most commercial SATCOM networks rely on proprietary network management systems. This paper addresses use of a Manager of Manager architecture, supported by local intelligent agents, for managing a global “mixture of media” SATCOM network with multiple access domains. Emphasis is on managing cross-domain roaming, where challenges include terminal mobility, waveform switching, adaptive modulation/coding, antenna beam handover, terminal reconfiguration, service transfer, along with associated management and control of multiple service providers in realtime. Policy planning and enforcement coordination between subsidiary network managers and element managers is discussed.

Directional TDMA Networking without External Time and Position References

The current generation of Directional Mobile Ad Hoc Networking waveforms relies on accurate, universal, externally generated time and position references that must be available at all nodes. Although in principle the waveforms can use any such references, satellite navigation systems such as GPS are the practical source of these universal time and position references. This reliance on GPS is an impediment to the adoption of DMANETs for some important applications. Methods for eliminating the reliance on external references have been investigated and the performance of the selected techniques has been assessed by simulation. A preliminary DMANET waveform architecture is described that only requires information locally available at each node.