Cognitive RF Systems and EM Fratricide Part II

Transcription

1 Cognitive RF Systems and EM Fratricide Part II Gerard T. Capraro Capraro Technologies, Inc., 401 Herkimer Road, Utica, NY USA Abstract The United States Department of Defense and researchers throughout the world have been addressing the overcrowding of the radio frequency (RF) spectrum. When the frequency spectrum is measured over time, technologists have shown that the spectrum is underutilized. This has led to numerous studies concerning cognitive radios, networks, and radar systems to intelligently choose frequencies, waveform parameters, antenna beam patterns, etc. to operate with conventional receivers without causing electromagnetic (EM) fratricide. In many of these studies there is an inherent assumption that the cognitive system knows when and where the fratricide occurs. In a previous paper [11] we presented two approaches for determining if a cognitive solution is causing EMI in nearby military conventional receivers. In this paper we extend the solution to include commercial transceivers along with military and describe a solution of how a radar and communication transceivers can cooperatively share the spectrum and reduce EM interference. 1.0 Introduction The radio frequency (RF) spectrum is crowded and more space is needed for wireless internet access, cell phone communications, and for military and civilian usage. The US Congress passed a bill to open up more spectra [1] to auction off RF frequencies belonging to the television broadcast industries. To slow down the need for more frequencies and to reduce the number of cell towers to accommodate the growing number of mobile phones, the industry is deploying microcell, picocell and femto cell technologies. However, these initiatives alone will not solve spectrum crowding. When the frequency spectrum is measured over time, technologists have shown that the spectrum is underutilized. Recognizing this, there have been numerous research projects funded by the US Department of Defense (DOD). The Defense Agency Research Project Agency (DARPA) has probably funded the most projects in this area. Through this research, we now have two distinct users defined as the primary user (PU) (i.e. those who own the license for the frequency range) and the cognitive user (CU) (i.e. those users trying to share the spectra either by using broadband signals or sampling the spectra in time and transmitting when the PU is not transmitting). Most significant projects in this area include the DARPA XG program and the Wireless Network after Next (WNaN) program. In addition to these efforts, there has been a move to apply Cognitive Radio (CR) technologies to the radar domain (Cognitive Radar efforts) and radio networks. Some of these systems sample the spectrum and transmit if no one else is transmitting at any given frequency. This approach can cause electromagnetic interference (EMI) in nearby receivers. Many people have recognized this problem and have addressed it in many different ways [2-6]. Many of their solutions inherently assume they know information about the victim receiver, but do not address how this information is obtained. Our last paper [11] addressed this issue. We presented two approaches for determining if a cognitive solution is causing EMI in nearby military receivers. Since the publication [11] DARPA released a request for proposal (RFP) [16] asking for approaches to solving the EMI problem through techniques that would cooperatively share the spectrum between military radar systems and military and civilian communication systems, e.g. mobile cell communications. This paper will briefly review the results of the previous paper and present an approach of how radar and communication systems can cooperatively share the spectrum and reduce EMI. Section 2 briefly reviews cognitive radio and radar technologies. Section 3 defines the problem we are addressing. Section 4 reviews two documented solutions [11] and presents a new solution based on cooperatively sharing of the RF spectrum. Section 5 provides a summary and conclusions. 2.0 Some Cognitive Efforts Next Generation (XG) Program The XG (next Generation Communications) program is developing an architecture that will open up the spectrum for more use by first sensing and then using unused portions of the spectrum.

2 Figure 1 from [7] is a logical functional diagram of the concept of operations of XG s policy-agile spectrum user, which uses a computer understandable spectrum policy capability. The major components are the Sensor (which senses the environment for determining its availability), Radio (the communications device that can dynamically change its emission and reception characteristics), Policy Reasoner (manages spectrum policy information), and System Strategy Reasoner (manages the multiple radios on a platform). similar goals to the XG program, is the Cognitive Radio [9]. Its objectives are to efficiently utilize the radio frequency (RF) spectrum and to provide reliable communications at all times. A basic cognitive cycle view of the radio is illustrated in Figure 2. A general overview and projections of the Cognitive Radio in our society can be found in [10]. Figure 1.0 Policy-Agile Operation of XG Spectrum- Agile Radio The last two components are of particular interest in that they utilize Semantic Web technologies. Operating a radio in different parts of the world requires that radios abide by the policies in the area where they are located. The XG program has developed its own XG policy language (XGPL) which uses OWL as its standard representation and will be implemented within the Policy Reasoner. The Wireless after Next (WNaN) The WNaN being performed by Raytheon BBN Technologies and funded by DARPA [8] is developing a scalable, adaptive, ad hoc network capability that will provide reliable communications to the military. The basic ingredients of their design are composed of a Dynamic Spectrum Address capability based upon the XG program. It also has 4 multiple transceivers and a disruptive tolerant networking (DTN) capability. The four transceivers provide fault tolerance and allows the system to pick the best channel for communications. The DTN capability allows the nodes to store packets temporarily during link outages. The WNaN also has content based access that allows users to query the network to find information and allow the system to store critical data at locations to minimize time and bandwidth. Figure 2.0 Basic Cognitive Cycle Cognitive Radar Interest in cognitive radar is growing in the radar community. Figure 3 describes a cognitive radar that is primarily concerned with the tracking stages of a radar [12]. In Figure 4 a cognitive radar architecture is shown from the first textbook written on this subject [13]. The commonality of these designs are the feedback loop between the transmitter and receiver, use of outside sources of information, and the implementation of a learning process. Figure 3.0 A Cognitive Tracking Architecture Cognitive Radio Another effort related to communications, and having Figure 4.0 Another Cognitive Radar Architecture

3 3.0 Problem Definition In all of the above programs and many others, the CU chooses which frequency to transmit using frequency policy rules based upon location and whether someone is currently transmitting within the range of interest. There are at least three issues with this approach. The first is related to the sensing of the environment. What happens if a nearby receiver is not transmitting but is waiting to receive a signal at frequency f1, for example a bistatic radar receiver or an electronic warfare receiver? They don t transmit, they just receive. The second issue relates to the following scenario. Let us assume that one decides to transmit broadband signals below the sensitivity levels of any nearby receivers. As the number of CR increases, the signals within a nearby receiver s passband may exceed the noise floor and interfere with the performance of the receiver [14]. The third issue occurs when a CR decides to transmit at a particular frequency because there are no signals present. The chosen frequency is based upon a linear relationship between the frequency chosen and the sensed environment. The decision policy does not take into effect the nonlinearities between the chosen frequency and other nearby frequencies which can mix nonlinearly and cause receiver intermodulation or mix within the receiver s frontend and cause spurious responses. Most electromagnetic interference (EMI) situations are nonlinear. EM Compatibility Paradigm Shift EM fratricide is the situation where we degrade the performance of our own system(s) with our own system(s), e.g. an onboard radar s energy is received by an onboard communication receiver and that degrades the receiver s performance. This is a serious problem, since there are multiple sensor and communication systems onboard platforms. Military weapon systems are engineered to prevent such phenomena between hardware located in close proximity. The military has standards for describing how to build and test hardware for EMC, and how to test weapon system platforms for EMC, e.g. Military Standards 461E and 464. In the 1970 s and 1980 s the DOD developed EMC prediction tools to assess the EMC of its weapon systems. Using these software tools to perform EM measurements in the 1970s was a major paradigm shift for the EMC community. Just as we needed a change by using software tools to assess a system s EMC in the 1970s, we now need to rethink how to build complex systems that employ waveform diversity and some of the proposed XG and cognitive radio and radar spectrum management concepts. Whereas in the 1970s we required software tools to predict where to hone our measurements, we now need to use software to help determine when EMI may occur in real time, and manage the EM spectrum while the platform or radio/radar increases its total performance. 4.0 Potential Solutions To solve the EM fratricide issues discussed above some people are looking to change the beam pattern of the transmitter so that the power coupled to a victim receiver is reduced, some wish to change the transmitted signal s polarization, and of course, there is the attenuation gained by employing orthogonal waveforms. All these solutions help reduce the amount of degradation caused to a friendly receiver. However, these techniques inherently are assuming that one knows that the receiver is being degraded. How would a cognitive radar, radio or a WNaN know about the receiver? There are currently three scenarios where one can implement a capability to solve the fratricide issue. One is on a single platform such as an aircraft, ship, or a complex weapon system where multiple conventional and cognitive EM equipment reside. The second scenario is concerned with WNaN where we propose to extend its capability and add a gateway to communicate with non cognitive radios as developed under another DARPA program. The EMC paradigm shift for both scenarios requires that the equipment report to a node that is managing the EMC of the platform or the total network. A discussion of these two scenarios can be found in [11]. The third approach is the cooperatively sharing of the RF spectrum which may require modifications of all RF transceivers. The third approach is concerned with the growing demand for complimentary coexistence between radar and communications devices that concurrently and cooperatively operate in shared spectrum. Over the past several years, DARPA and other DOD agencies have initiated numerous research projects in dynamic spectrum sharing among various military and commercial systems. Some of these approaches were discussed above. However, these spectrum sharing approaches, which relied upon spectrum sensing, did not map the presence of recently non-radiating nodes nearby. They do not implement cooperative techniques amongst those sharing the spectrum, nor have they taken into consideration the benefits possible through anticipating the near-term actions of other users. The DARPA RFP [16] addresses these issues, and requests solutions having radar and

4 communications devices share information on EMI and how to best remedy the situation in near real time. Advances in wireless networking and information management technology have resulted in the proliferation of communications devices (e.g., smartphones) and mobile ad hoc networks (MANET). While waveform diversity and advanced processing has led to the deployment of multifunction wideband radars and sophisticated RF sensor systems on unattended aerial vehicles (UAV), employing embedded computing and geographical/global positioning systems (GPS). Each of these developments requires additional spectrum to achieve mission goals. This RFP addressed three real world scenarios, the interaction of radar with 1.) small cell broadband (SCB), 2) Wi-Fi hot spots for the commercial world and, 3.) MANET systems for the military world. Operations in an electronic countermeasures environment must also be considered. Figure 5 illustrates the problem. The number of cell phones is growing at a faster rate than predicted just ten years ago. In order for the industry to provide good coverage in dense environments, including urban areas, while minimizing the cost of additional cell towers, SCB technology [17] is being pursued to exploit the Internet. On the military side, the development of MANET allows communications devices to be mobile and yet be able to communicate long distances without substantially increasing power and without building expensive communications infrastructure. Figure 5 presents a scenario with radar having the higher priority (primary) and communications as secondary. The first question the radar cognitive processor needs to answer is whether the radar is being interfered with by a MANET, SCB, heterogeneous transmitters/receivers (T/R) or Wi-Fi, and thus requires receivers sensing these bands. In addition, it must quantify the impact of this interference. This cannot be accomplished via noise power measurements alone, and must be signal structure sensitive. The radar requires cognitive embedded calibration such as the insertion of known signals (non-interfering false targets) in the receiver front-end. In doing so, the radar will correlate how well it detects these false targets against other signals that it receives from nearby emitters. If a particular communications system is identified as causing interference, then the radar can embed an encrypted message in its next transmission alerting the communications system to the impact, and the two systems can negotiate how to eliminate the EMI. Alternatively, the communications system will communicate to the radar, its position, power output, location, modulation, frequency, antenna pattern, etc. This will help the radar cognitive processor determine which system is causing the EMI. In addition the radar will also transmit similar information about its transmissions so that the communications systems can use this information when they have a higher priority and the radar may be causing EMI. When communications devices are primary, they will broadcast while receiving specifics about the radar transmissions. The broadband routers and other heterogeneous T/R devices will require a cognitive capability and receiver that can detect encrypted radar messages, or, alternatively, receive notification messages from other T/Rs. In this manner they can assess whether they are being interfered with by the radar systems. This may be accomplished by correlating the radar transmissions with the packet errors detected during communications. Correlations of excessive packet errors along with distance, timing, antenna characteristics, terrain, and power rules will allow the communication systems to determine how the radar impacts performance. The T/R nodes can communicate to the radar, via distributed control protocols, describing its EMI and jointly, the two cognitive processors will alleviate the EMI. Figure 5 Cooperative Spectrum Sharing Each T/R node will have a software architecture similar to that shown in Figure 6. Each node should have mission goals and be able to set priorities. We have to establish performance metrics to know how well each node is functioning. We will use databases and knowledge bases to assess and maintain a node s status, and ontologies to describe the rules of spectrum operation in different theaters. Classification is

5 necessary for radar in defining clutter statistics and in communication nodes to help define multipath and propagation loss. To assess how well a node is performing, we implement a series of tests. Background analysis is a required capability for computing propagation losses, multipath, clutter statistics, etc. and they will use, for example, Land Use Land Cover (LULC) and Digital Elevation Map (DEM) databases. Auxiliary information is required to obtain information such as jammer locations. Figure 6 Cognitive Node Software Architecture 5.0 Summary/Conclusions We have briefly reviewed two approaches for solving EM fratricide where cognitive systems are functioning near conventional T/R [11]. A new approach is being investigated by DARPA where communications and radar systems will cooperatively share the spectrum. We have described the problem and an approach where each T/R node will have a cognitive software architecture as shown in Figure 6. This capability will be embedded in military/commercial radar and radios e.g. MANET, cell towers, WiFi routers, and SCB routers. There is a lot of work that needs to be done in developing these cognitive nodes and how they will cooperatively work together to solve EMI in near real time, learn from their discoveries and pass this knowledge onto other T/R nodes so they can learn from each other. References [1] [2] P. Popovski, Y. Hiroyuki, K. Nishimori, R. Di Taranto, and R. Prasad, Opportunistic Interference Cancellation in Cognitive Radio Systems, 2nd IEEE International Symposium on New Frontiers in Dynamic Spectrum Access Networks, [3] O. Ozdemir, E. Masazade, C. Mohan, P. Varshney, I. Kasperovich, R. Loe, A. Drozd, and S. Reichhart, Spectrum Shaping Challenges in Dynamic Spectrum Access Networks with Transmission Hyperspace, Waveform Diversity and Design Conference, January 2012 [4] Y. Fei and Z. Wu, An Interference Cancellation Scheme for Cognitive Radio Network, 6th International Conference on Wireless Communications Networking and Mobile Computing, Sept [5] V. Chakravarthy, Z. Wu, A. Shaw, M. Temple, R. Kannan, and F. Garber, A general Overlay/Underlay analytic Expression Representing Cognitive Radio Waveform, Waveform Diversity and Design Conference, June 2007 [6] E. Beadle, A. Micheals and J. Schroeder, New Alternatives for Interference Tolerant Waveforms Hosted on a Software Programmable Multi-Mission Platform, Waveform Diversity and Design Conference, January 2012 [7] D. Elenius, G. Denker, andd. Wilkins, XG policy Architecture, ICS TR , SRI Project No , Contract No. FA C-0230, April 2007 [8] J. Redi and R. Ramanathan, The DARPA WNaN Network Architecture, The 2011 Military Communications Conference, November 2011 [9] S. Haykin, Cognitive Radio: Brain-Empowered Wireless Communications, IEEE Journal on Selected Areas in Communications, vol. 23, no. 2, pp , February 2005 [10] Ashley, S. Cognitive Radio, Scientific American, ID=1&articleID=000C7B F6- A B7F0000, February 20, [11] G. Capraro and I. Bradaric, Cognitive RF Systems and EM Fratricide Proceedings of the 2012 International Conference on Artificial Intelligence, Volume I, pp , July 16-19, 2012 [12] S. Haykin, Cognitive Radar, IEEE Signal Processing Magazine, pp , January 2006 [13] J. Guerci, Cognitive Radar, Artech House, 2010 [14] I. Bradaric, G. Capraro, and D. Weiner, Ultra Wide Band (UWB) Iterference Assessment And Mitigation Studies, AFRL-SN-RS-TR , February 2006, [15] H. Keyton, Networks: Adapting To Uncertainty, DARPA 50Years of Bridging the Gap, [16]DARPA-BAA m&id=8e85f738e53747b502b4b9c3732c2e54&tab=co re&_cview=1 [17]