By Dean ~ August 21st, 2012. Filed under: Modeling, SDR & SCA, Tips & Tricks.
Mobile Ad-Hoc Networks (MANETS) are self-forming, self-healing mesh networks that enable reliable high-bandwidth communications in dynamic environments. While their benefits are obvious, the challenges encountered in implementing and testing these complex networks can be extremely daunting. Sophisticated simulation is the most effective way of identifying and addressing issues early in the design process, before costly mistakes become embedded in the implementation.
With Foresight’s Resource Aware Modeling & Simulation (RAMS) and OPNET Modeler, we have the technology to design MANETs that meet their expected operational parameters, or at least to understand why that might not be possible. In the latter case, we might have to go back and revisit the requirements for the system.
In the scenario discussed here, the OPNET model addresses the network behavior and includes the behavior of the waveform running on all of the nodes. Meanwhile, the Foresight model is focused on the performance of the waveform on one, carefully chosen, node given the resource constraints of the radio platform. In this combination, the OPNET network simulation generates the stimulus for the Foresight node model. The Foresight simulation, in turn, generates real-world waveform/node characterization data that can be back-annotated to the OPNET model as latency approximations. These latency annotations allow the OPNET network model to provide a more accurate representation that reflects the impact of radio platform performance. We’ll explore how this all comes together in the rest of this post.
The difficulty in developing a complete picture of MANET operation stems from the need for insight at two vastly different levels of abstraction. Teams must understand both the specifics of the underlying hardware/software systems and the dynamics of the wireless links. While Foresight’s tool can certainly model both aspects of the system, some teams may be more comfortable with OPNET for modeling the wireless network. They may even already have an existing OPNET model of the network. Fortunately, for those situations, we can employ this pair of best in class tools to explore the entire universe of MANET issues.
When used together, the software from Foresight and OPNET integrates smoothly to form a powerful combination that gives design teams unprecedented visibility into the projected reliability and bandwidth of their MANET. OPNET Modeler provides an excellent representation of the dynamics of the wireless links, while Foresight’s RAMS technology supports a clear understanding of the constraints and performance of the underlying hardware / software systems. Together, these technologically advanced tools give teams complete insight into the performance, behavior and potential issues of a proposed MANET design.
Let’s explore this tool-set in a bit more detail. Obviously, the resource demands and traffic encountered in the field will drive the workload for specific nodes. We can gain an understanding of these complex dynamics of the network by developing high fidelity protocol models with OPNET. These models may be either behavioral or shared code. Both types allow us to gain detailed insight into the operation of the network protocols in simulated space-time.
As we’re developing the OPNET models, we identify data that will be exported to Foresight. Our objective is to select events and data objects that can be used to drive the execution of the node simulation in Foresight. XML can provide a convenient format for converting between OPNET protocol interface messages and the information that will drive Foresight’s analysis. Other options for exchange are also available.
Of course, both network bandwidth and detailed node performance are primary determinants of the amount of information that can be reliably exchanged over the network. Consequently, to fully grasp the complex interactions within the network, we also need to consider a lower level of abstraction — the processing, throughput and behavior of individual nodes. We accomplish this through exploration of the Foresight RAMS model, which enables us to understand the performance and interactions of protocol stack layers on the target hardware / software systems. These constructs represent both:
- the Behavior — what the system does, in this case the waveform that we’re investigating
- the Platform — the collection of software services, middleware, RTOS, processors, memory and busses that implement the lower levels of the system
The RAMS model provides a detailed view into each node’s performance and bottlenecks.
As the Foresight model executes, timed stimuli derived from the OPNET network simulation drive the inputs to the Foresight model. These may include transmit and receive data packets, control events, and any other internal events for which the Foresight behavioral model cannot directly provide the network traffic context. This allows us to analyze the behavior of a node, or set of nodes, within the context of network traffic produced by the OPNET simulation.
Within the Foresight model, point-to-point latency across waveform software components is logged as well as detailed resource behavior. From this information, a broad range of data can be derived, including:
- Latency and jitter per waveform software component over time
- Latency and depth behavior of queues over time
- Resource utilization over time
- Resource utilization by client over time
- Run time vs. wait time by client over time (This can be used to identify thread prioritization problems.)
The latency data from the Foresight simulation can then be annotated back into the OPNET model to improve the accuracy of the network simulation. The additional latency information allows us to measure overall network performance with the realities of the platform taken into account.
MANETs are foundation technology for many networks of the future, but they’re beasts to get right. Although Foresight RAMS could be used to model both the network links and the communicating nodes, for those teams that have an existing investment in OPNET, using the powerful combination of Foresight and OPNET will go a long way towards ensuring that your MANET meets its performance and cost objectives.
If you’re interested in learning more, please consider the following resources:
- Whitepaper (PDF): Resource Aware Modeling & Simulation
- Whitepaper (PDF): Comparison of Foresight and Mil3 OPNET
- Performance Engineering Blog: Using RAMS For Software Defined Radio Design