Michael J. Markowski

Fully distributed wireless MAC transmission of real-time data

As ad hoc wireless networks become more common, the need for real time services on these networks is emerging. However little work has focused on supporting real time requirements in such a demanding environment where the medium of transmission is noisy and protocols must be fully decentralized. Addressing this, we present a MAC protocol, preemptive in nature, that supports the transmission of hard, soft and non real time data. To guarantee stability even in overloaded conditions and still offer support for deadlines, we use splitting protocols. A protocol model is presented and analyzed, showing its performance under a wide range of offered loads with various fractions of real time types. We show that the protocol is easily implemented without making unreasonable hardware demands and so could be used as the basis for building real time services in an ad hoc wireless network.

Wireless MAC protocols for real-time battlefield communications

Addressing the problem of timely packet transmission in a wireless soft real-time system such as one would find on the battlefield we present five splitting protocols that take packet deadlines into account. Three are blocked access and two are free access algorithms. Mathematical models of the algorithms are developed, and compared with simulations. We show, as expected, that the blocked access algorithms usually offer higher success rates than the free access versions. Of the two best performing blocked access protocols, performance differences are slight on a lightly loaded channel. Under heavy load, however, one is shown to have better performance, and thus would be the best choice for implementation.

Fully distributed wireless transmission of heterogeneous real-time data

Using wireless systems to transport real-time data is becoming increasingly common as wireless networks themselves are used more. However, in most instances, real-time data is generated by applications and ultimately transmitted by a MAC layer protocol with no support for deadlines. We describe a MAC layer protocol that transmits data taking into consideration its real-time properties. This would appear to be the first wireless MAC protocol offering support for hard, soft, and non-real-time data in a preemptive manner. Numerical results of the mathematical model (not presented here) are graphed and evaluated.

Evaluation of wireless soft real-time protocols

Communication between current military real-time systems and future interconnection of general purpose, embedded real-time systems will often require wireless communications. However, there has been little work undertaken to offer support for real-time applications on wireless networks. We present and evaluate three protocols; variations of two published protocols by Paterakis and Gallager as well as our new one, the Sliding Partition (SP) collision resolution algorithm (CRA). In a real-time setting, the modified Gallager CRA consistently performs worst of the three we consider. We observe that when the deadline range is small, the Sliding Partition CRA performs best. When the deadline range is large, however, the Paterakis CRA performs slightly better than the SP CRA. Both analytic and simulation results are obtained to study the maximum input traffic rates that can be sustained for various laxities, delay bounds, and message loss rates.

Modeling mobile network connectivity in the presence of jamming

The Network Connectivity Analysis Model (NCAM) is a general purpose radio frequency (RF) event-based simulator that addresses the physical layer of the International Organization for Standards (IOS) Open System Interconnect (OSI) 7-layer model for wireless communication systems. With a modular design hierarchically organized to emulate an actual physical test setup, NCAM provides the U.S. Army with an extremely flexible modeling and simulation capability that could predict the impact of RF interference on mobile tactical communication networks. NCAM can simulate any number of radios or jammers in ground-based platforms. This paper describes the algorithms of each module in the model and the capabilities and features of its event-based simulation that supports dynamic state changes such as antenna swap, antenna height adjustment, radio on/off, etc. It can also pause, save, and resume simulation at a later time, for classified runs. Verified and validated by the Communication-Electronics Research, Development and Engineering Center (CERDEC), NCAM has been used to support tactical communication networks analysis.

Real-time wireless communication using splitting protocols

Addressing the problem of packet transmission in a wireless soft real-time system, we present five splitting protocols that take packet deadlines into account. We show, as in the case of non-real-time splitting algorithms, that blocked access versions offer higher success rates than free access ones. Of the two best performing blocked access protocols, the performance under moderate to heavy loads further shows the superiority of the sliding partition collision resolution algorithm (CRA) over the two cell CRA.

Simulating Relaxed Pedestrian Behavior

A model of pedestrian movement in a comfortable environment is presented. “Comfortable” and “relaxed” here mean that traffic flow is not overly restricted by environment or congestion and in addition is not related to safety measures, such as evacuation. The algorithmic model uses a layered approach to model human movement in which rules and behaviors are created to simulate group interactions. Learning and memory are modeled so that simulated pedestrians are like humans in ways that have an effect on congestion, and software has been developed that implements the model. The model was calibrated by using widely accepted data and validated against observed data in a shopping mall. Results from the simulator show that the model produces reliable results for situations in which the modeled behavior is typical. Proof-of-concept software implementation shows the utility of the model and how it can be used to ease and improve design of pedestrian areas.