A.R.K. Sastry

A performance comparison of control policies for slotted Aloha frequency-hopped multiple access systems

A performance analysis of slotted Aloha with frequency-hopped code-division multiple access is presented, with emphasis on the effects of the retransmission control policy. Static and dynamic control policies are considered. Performance using static control is shown to be highly sensitive to the relationship between the packet generation process and the chosen retransmission probability. Dynamic control, on the other hand, is very robust in that it provides high throughput over wide ranges of arrival rates. Delays are generally quite small when system parameters are properly set. A simple rule for computing the retransmission probability as a function of the backlog is provided and gives a result very close to the analytically derived optimal value. Numerical performance evaluation is given for the case of a network of 250 users and a spread factor (number of hopping positions) of 100, where the other system parameters are varied.<<ETX>>

Code division multiple access with code assignments using information classes

The authors propose a new method of code allocation in code division multiple access, based on information classes. Transmitters choose spreading codes based on the type of message they are transmitting, and receivers listen to codes corresponding to the type of message they are interested in. The performance of this approach can exceed that of both receiver-directed codes or broadcast codes in environments where there are several multicast groups (in which a given message is of interest to more than one node), or where it is difficult to know what types of messages (out of a fixed set of message types) a receiver is interested in at any particular time. Performance is further enhanced when receivers can simultaneously attempt to correlate multiple spreading codes.<<ETX>>

Application of knowledge-based network management techniques for packet radio networks

The authors developed a preliminary version of a knowledge-based model for network management and reconfiguration using blackboard techniques and have applied to it packet radio networks. The analysis is concerned with developing procedures for evaluation of candidate recovery/reconfiguration methodologies and techniques for fault isolation and related monitoring functions. As an initial step, the generic blackboard was chosen as the artificial intelligence environment to develop the management tools and interlink it to a packet radio network simulator that was used as a testbed network to be controlled and monitored. The details of the interaction of the management environment and the packet radio simulator as implemented in the model so far, and present numerical results obtained through the execution of some preliminary rules are described.<<ETX>>

A simulation model for evaluation of distributed processing in multi-hop packet radio networks

The authors describe a simulation model developed to aid in the evaluation of distributed processing scenario in a highly dynamic environment such as a multihop packet radio network. The simulation model has a set of identical packet radio nodes, employs spread spectrum random access protocols, and has an error model that includes the effects of interference from concurrent transmission as well as that induced by an on-off partial band jammer. An abstract description of hierarchical primary and subtask structures has been devised wherein a distributed task is described by a script that specifies processing time, sequences of subtasks, and lengths of request and response messages. The features of multihop packet radio networks are described that are relevant to the effort. The details of the network simulation, including the transport features are given. The distributed task scenario description is outlined. Some typical numerical results obtained from simulations are presented.<<ETX>>

Dynamic allocation of processing capacity in a multiband radio

A configurable multiband radio (MBR) with the ability to generate multiple simultaneous waveforms from HF to SHF will play a significant role in future tactical military communications and in facilitating interoperability on the global grid communications infrastructure. The authors initiated work on processing strategies within an MBR with multiple processors. Traffic requirements and processing needs can be expected to vary widely for different channels from HF to SHF. Thus, rather than having dedicated processors, it will be advantageous to allocate processing power to each channel on demand. The authors describe a simulation model to understand the impact of shared processing in a MBR and related issues. They provide details on the simulation structure, the object comprising the simulation, and operation of the simulation. The simulation deals with sharing of MBR processing and RF assets among multiple simultaneous users within a network of such MBRs. A traffic model provides a variety of traffic scenarios including priority, point-to-point and multicast, voice or data media, and datagram, request/response and request/acknowledge protocols. Buffering on sending and receiving side is permitted. The authors conclude with some simulation results indicating tradeoffs between the number of processors and loss of messages due to processing demands, and the impact of task segmentation, buffering, and channel traffic levels.<<ETX>>

Knowledge-based management of multi-hop packet radio networks

The authors have developed a knowledge-based model of network management and applied it to multihop packet radio networks for fault diagnosis and transmission control. The model has two distinctive features: (1) a distributed control concept and (2) the use of the blackboard paradigm as a knowledge-based environment. The focus of the effort has been on diagnosing faults and other abnormal conditions and applying corrective actions. For transmission control, the authors have implemented a control policy based on observed channel occupancy. At the transport level, the messages are segmented/reassembled, and a facility has been created to allow nonuniform traffic among different source-destination pairs. The authors present a brief description of the software testbed consisting of a packet radio network simulation and the blackboard environment, describe the control rules that have been implemented, and summarize the results obtained.<<ETX>>

A simulation testbed for LPD networks with power control

There is a growing interest in low probability of detection (LPD) communication networks due to increased focus on tactical networks, and due to improved processing capabilities for spread spectrum waveforms. The probability of detection by an interceptor is affected by design decisions made throughout the hardware and protocol levels of the communication system. Because of the difficulty of analytically modeling the effects and interactions of system design choices made at different levels of the system upon the probability of detection, particularly in the networking context, the authors have developed a combined simulation and analytical model of a network of mobile communicating nodes operating in the presence of jammers and hostile communication interceptors such as power radiometers. They use the model to examine the efficacy of dynamic power control in LPD networks.<<ETX>>

A simulation model for performance evaluation of ISDN user part basic call control procedures

A preliminary integrated services digital network (ISDN) model is developed that incorporates basic features such as voice and data call generation, call setup for circuit-switched calls, packet-switched signaling, transmission of low-rate data on the signaling channel, and selectable number of nodes, link capacities, and connectivities. The model concentrates mainly on the basic call control procedures of the ISDN user part of the CCITT Signaling System No.7 in conjunction with D-channel protocols. The primary objective of the ISDN simulation model is to aid in understanding the performance of the ISDN protocols. In the design mode, it may be used to evaluate a number of potential architectures consisting of various message traffic/workload scenarios with different mix of voice and data traffic, resource additions and expansions and congestion handling. The model produces statistical measures of network utilization and call blockage that can be used to evaluate the effect of changes in network topology. routing algorithms, and traffic. The architecture and input/output formats of the model are described. Some numerical results from sample runs are given.<<ETX>>