 |
 |
 |
 |
| |
|||||
|
|
|
Most networks today have a fixed infrastructure with reliable, high-capacity links. As a result, the protocols developed to support user communication needs have assumed this type of network. What happens when the network infrastructure is dynamically changing, and the links are wireless with less capacity and are more prone to errors? This type of network is referred to as a mobile ad hoc network. ("Mobile" because the network's constituent elements are moving, and "ad hoc" because it does not depend on any fixed arrangement of those constituent elements). Such networks will be critical to future deployed military units, and several Army programs (such as the First Digitized Division) have requirements for this type of network. To support a user's need to send information to an intended recipient, a network must be able to route information from one user to another. Each element in a wireless network must support the job of routing by being able to do two important tasks: First, it must be able to establish two-way radio communication with its neighbors. Second, it must be able to use data exchanged with its neighbors to figure out how to send information to a user that is not its immediate neighbor. When the network is also mobile and ad hoc, all the radios can be moving, so neighbors can be changing. In addition, links between neighbors can be unidirectional (information can be sent and received in one direction but not the other) or asymmetric (for example, one direction has a higher capacity than the reverse). For the Army, the radio network also needs to support the different service requirements of applications, such as voice (time-sensitive) or bulk transfer (loss-sensitive), and possibly differing requirements between users (such as a general or a private). MITRE is investigating Army-specific mobile ad hoc issues, with a focus
on medium-term solutions. The main objective is to recommend advanced
radio network architectures and protocols to support the Army's First
Digitized Division and advanced radio developments at lower echelons.
Secondary objectives are to develop simulation models using MIL3's OPNET
Modeler package for current and future evaluations and to expedite development
of a Request for Comments (RFC) on ad hoc routing in the Internet Engineering
Task Force (IETF). MITRE is also recommending routing and channel access
protocols in tactical ad hoc radio networks, developing OPNET simulation
models of protocol alternatives, and establishing performance baselines
of these protocols for different tactical conditions. The project's simulations
and research are already being transitioned to other projects including
the Defense Advanced Research Project Agency's (DARPA's) Small Unit Operations
Situation Awareness System program and a MITRE project that has just begun
on satellite mobile mesh networks. For any type of transmission in a multi-node network, a transmitter must first gain access to the channel over which the information will be sent. Techniques to access the channel range from contention-based to reservation-based. Contention-based methods use packet-by-packet contention for the radio channel and can lead to conflicts between nodes wanting to transmit at the same time. Dynamic reservation-based methods reserve the channel for transmission of a packet or series of packets in a message or call. Contention-based channel access tends to provide the best delay and throughput characteristics when traffic consists of small packets with bursty interarrival times, while reservation-based channel access tends to provide the best performance when traffic consists of large messages or streams of packets in a call. MITRE's project is investigating channel access algorithms that combine random access with reservation-based methods to minimize node transmission conflicts to provide the best overall performance for mixed data and voice steam traffic. The algorithms include Multiple Access with Collision Avoidance, and distributed, dynamic Time Division Multiple Access techniques. We are evaluating these algorithms for a variety of tactical conditions. In addition to gaining access to another user, the network must figure out a path to that user. This is the function of the routing protocol. Routing protocols for mobile ad hoc networks generally fall into one of two categories: proactive or reactive. Proactive routing attempts to maintain routes to all destinations at all times, regardless of whether they are needed. To support this, the routing protocol propagates information updates about a network's topology or connectivity throughout the network. Information updates can be topology-driven, which are generated when connectivity in the network is detected; periodic, which generates connectivity information at fixed intervals; or both. Topology-driven updates provide optimal routes if network connectivity is stable, while periodic updates limit overhead deterministically at the expense of route optimality and responsiveness. These updates can be incremental (include only route changes) or full (include all route information). Traditional routing protocols such as Open Shortest Path First (OSPF) are examples of proactive routing protocols. An example of a proactive routing protocol tailored to the mobile ad hoc environment is the Wireless Routing Protocol (WRP). This protocol was developed under DARPA's GloMo program to make more efficient use of the bandwidth than traditional proactive routing protocols. For example, in WRP, routing updates are sent to neighbors only when something new is discovered or information is updated. These topology-driven updates are also incremental, including only information that has changed since the last update. WRP also includes several techniques to reduce the likelihood of routing loops, which can consume bandwidth and increase packet delays. However, as a result of these enhancements, WRP is much more complex than traditional routing protocols. In contrast, reactive or on-demand routing protocols determine routes only when there is data to send. If a route is unknown, the source node initiates a search to find one, which tends to cause a traffic surge as the query is propagated through the network. Nodes that receive the query and have a route to the requested destination respond to the query. In general, reactive protocols are primarily interested in finding any route to a destination, not necessarily the optimal route. Data sent in networks using reactive protocols do tend to suffer a delay during the search for a route. Under highly dynamic link conditions, reactive protocols are expected to generate less overhead and provide more reliable routing than proactive routing, but at the cost of finding the optimal route. Examples of reactive protocols include Temporally-Ordered Routing Algorithm (TORA) and Ad Hoc On Demand Distance Vector (AODV), which are among the routing protocols being considered by the IETF MANET Working Group. We are investigating the effectiveness of different routing protocols under varying conditions, such as link fluctuations (an indication of node mobility), node connectivity (an indication of how many neighbors a node has), loading, and bit-error rate. We conducted an initial quantitative evaluation of a reactive routing protocol (Corson and Ephrimedes, the predecessor to TORA) and of flooding (a routing algorithm where every incoming information packet not addressed to the given node is retransmitted and where no routing information is exchanged; this was used as a baseline). In this evaluation we found that the reactive protocol performed very well in dense connectivity for a wide range of traffic loading, but was less effective for sparse connectivity. This was because the protocol was able to use alternative routes it had collected in the initial query, when the primary route was no longer an option. Flooding was effective only for networks with sparse connectivity and low loads; with high connectivity or higher loads, the network was saturated. We are now undertaking a more comprehensive evaluation of several additional routing protocols, including some proactive routing algorithms tailored for the wireless environment (such as WRP) and reactive algorithms (such as TORA). In the current routing evaluation, network and end-to-end performance will be evaluated as a function of several factors, including traffic loading, network size, node connectivity, and link fluctuation rate. The results of this evaluation and the channel access evaluation will then be used to make recommendations on architectures and protocols to support advanced radio developments at lower echelons for the Army. For more information, please contact Nancy Shult using the employee directory. |
Solutions That Make a Difference.® |
|
|
To date, MITRE's team has conducted a broad qualitative assessment of
radio network architecture and protocol alternatives and has performed
an initial quantitative evaluation of routing and channel access alternatives.
Currently, we are conducting a broader evaluation of protocol alternatives
including routing alternatives from the DARPA GloMo program and the IETF's
Mobile Ad Hoc Network (MANET) working group. The rest of this article
gives a brief overview of routing and channel access protocol alternatives
and of MITRE's plans to analyze them.