B. Multi-beam Directional Antennas and Directional Networking
Traditionally, medium access control (MAC) protocols are designed for nodes, which are equipped with omni-directional antennas. Some disadvantages of using omni-directional antennas are poor data throughput, lower network and power efficiency due to interference resulting from the transmission of packets in undesired directions. Using directional antennas and the concept of sending packets in specific directions has a significant impact on spectrum efficiency and power consumption. Using multiple RF chains at each antenna element, separate beams can be formed simultaneously. The use of beamforming directional antennas capable of adaptively configuring multiple narrow beams and nulls enables (i) multiple simultaneous directional transmissions, (ii) spatial isolation and frequency reuse, ( i i i ) LPD/LPI qualities since transmissions are not broadcast in an omni-directional manner for anyone to hear/intercept, (iv) large data rates, (v) long transmission ranges, (vi) graceful degradation with failure and combat damage, (vii) interference avoidance, and (viii) antijam capabilities for contested environments by null formation in the direction of jammers and other undesired emitters.
However, there are challenges associated with the use of directional antennas that require more investigation. Using a directional antenna (i) requires the design of neighbor and topology discovery techniques for mobile nodes, (ii) introduces new hidden node problem due to node deafness, and (iii) requires distributed scheduling schemes to avoid performance degradation due to hidden nodes. Since antenna direction also impacts the routing path, a cross-layer design among routing-MAC-PHY is required.
Most of the published literature on directional antennas assumes a selectable main beam gain and no side lobe interference irrespective of beam width and number of beams, assumes non-overlapping beams, ignores range extension due to different beams widths, and does not consider terrain effects or contested environments. Riverbed Modeler is grounded in the math and physics of antenna and channel propagation. This makes the model results one step closer to reality because Riverbed Modeler prevents users from arbitrarily selecting the main-beam gain while ignoring the side lobe interference. Likewise, Riverbed Modeler correctly calculates the impact of beam width and number of beams on the transmission range, allows for importing a user-defined antenna gain table, and incorporates terrain and other propagation effects. Through preliminary comparative testing, AFRL/RI, in collaboration with San Diego State University (SDSU), has found that Riverbed Modeler simulations with directional antennas are far more realistic than the Matlab and ns-2 based simulations.
While Riverbed Modeler allows the use of directional antennas at the PHY layer, the implemented MAC protocols use omni-directional antennas. Consequently, the implementation of directional protocols required significant modifications to the existing Riverbed Modeler code. For the IEEE 802.11 MAC layer protocol, omnidirectional RTS/CTS/Data/ACK were modified to omnidirectional (or directional) RTS, omni-directional (or nearly omni-directional) CTS, and directional Data/ACK. This required writing additional code in the MAC process model and including cross-layer interaction between the PHY and MAC processes that did not exist in the standard 802.11 models. Further, the use of a multi-beam antenna equipped node required additional code for extending the existing omni-directional (or single beam) nodes in Riverbed Modeler. Since multi-beam nodes can simultaneously transmit and/or receive data on each of their beams, concurrent packet transmissions and receptions have been implemented, requiring additional code for simultaneous (or synchronized) RTS, CTS, Data and ACK, including addressing node back-off issues and alternate packet transmission-reception in relay nodes. The node model of a multi-beam antenna with four simultaneous beams and the multi-beam directional MAC is shown in Figure 5.
Currently, Riverbed Modeler uses a video data model where a fixed number of packets per second (corresponding to a given video bit rate and packet size) are assumed. Additional code was written to transmit H.264/AVC video bitstreams. H.264/AVC video has a variable number of packets, varying packet sizes, and a packet loss distortion value in each unit time. This enables the examination/ comparison of performances of different protocols and network topologies for the compressed full motion videos.
In the future, Riverbed Modeler will be used to implement the HMAC cross-layer protocol which allows for simultaneous transmission and reception of multiple packet on different beams, to design a MAC scheme for concurrent transmission and reception using nodes equipped with multi-beam antennas, to develop a new reactive routing protocol for finding multiple paths between source-destination pairs using multi-beam antennas in wireless mesh network architectures for comparing the new protocol’s performance with existing Riverbed Modeler protocols, to design a new MAC protocol to study the performance of TCP for long-distance links, and to examine protocol performance with respect to mission effectiveness in AF relevant environments containing high speed links, long distance links, interference, high node mobility, and jammers.
Conclusions
This paper has presented two efforts in networking MS&A. The discussion began with the basic theory, continued through the current implementations, and finished with planned future steps. As noted, planned laboratory and field tests will utilize this modeled performance analysis to reduce costs and increase to scope of available testing. The major components of a joint routing and spectrum allocation algorithm, ROSA, were presented, as well as current test results demonstrating that ROSA possesses high throughput, low delay, and fair bandwidth allocations in dynamic, ad-hoc networks., Finally, this paper presented the initial steps for developing a multi-beam directional antenna and networking capability in Riverbed Modeler.
References
[1] http://www.riverbed.com/products/steelcentral/steelcentral-riverbed-modeler.html
[2] L. Ding, T. Melodia, S. Batalama, j. Matyjas and M. Medley, “Cross- Layer Routing and Dynamic Spectrum Allocation in Cognitive Radio Ad Hoc Networks”, in IEEE Transactions on Vehicular Technology, vol. 59, no. 4, May 2010, pp. 1969-1979.
RELEASE STATEMENT
Distribution Statement A: Approved for Public Release; distribution unlimited: 88ABW-2015-4795