Electronic Engineer Essays

Electronic Engineer Essays Electronic Engineer Essay Electronic Engineer Essay International Conference on Modelling and Simulation Path Loss Effect on Energy Consumption in a WSN ? Krishna Doddapaneni,? Enver Ever,? Orhan Gemikonakli,†  Ivano Malavolta,? Leonardo Mostarda,†  Henry Muccini Communications Department, School of Engineering Information Sciences, Middlesex University, UK Email:{k. doddapaneni, e. ever, o. gemikonakli, l. mostarda}@mdx. ac. uk †  Dipartimento Di Informatica De L’Aquila Via Vetoio L’Aquila 67100, Italy Email:{ivano. malavolta,henry. muccini}@univaq. it ? ComputerAbstract- Energy consumption of nodes is a crucial factor that constrains the networks life time for Wireless Sensor Networks (WSNs). WSNs are composed of small sensors equipped with lowpower devices, and have limited battery power supply. The main concern in existing architectural and optimisation studies is to prolong the network lifetime. The lifetime of the sensor nodes is affected by different components such as the microprocessor, the sensing module and the wireless transmitter/receiver. The existing works mainly consider these components to decide on best deployment, topology, protocols and so on.Recent studies have also considered the monitoring and evaluation of the path loss caused by environmental factors. Path loss is always considered in isolation from the higher layers such as application and network. It is necessary to combine path loss computations used in physical layer, with information from upper layers such as application layer for a more realistic evaluation. In this paper, a simulation-based study is presented that uses path-loss model and application layer information in order to predict the network lifetime. Physical environment is considered as well.We show that when path-loss is introduced, increasing the transmission power is needed to reduce the amount of packets lost. This presents a tradeoff between the residual energy and the successful transmission rate when more realistic settings are employed for simulation. It is a challenging task to optimise the transmission power of WSNs, in presence of path loss, because although increasing the transmission power reduces the residual energy, it also reduces the number of retransmissions required. Index Terms- attenuation; path loss; wireless sensor networks; energy consumed; life time Evaluation tools Analytical modelingSimulators Real Deployment Test Beds Fig. 1. Performance evaluation methods I. I NTRODUCTION Recent advances in wireless communications and electronics have enabled the development of wireless sensor networks (WSNs), which comprise many low cost, low power, and multifunctional sensor nodes to accomplish certain sensing tasks in an intelligent manner. A sensor network is a special type of network which generally consists of a data acquisition system and a data distribution system. The unique characteristics of WSNs in terms of data collection and energy constrains, separate them from other communication networks.In Figure 1 we show the most common techniques for performance evaluation that are analytical modelling, simulation and benchmarking. The existing studies consider benchmarking in form of test beds and measurements for real deployment. The energy constrains of WSNs, limits their processing capabilities and communication. Therefore, using one of these performance evaluation methods, and analysis of deployment and management of such complex systems is a challenging task [1]. Due to inherent complexity and diverse nature of WSNs (dynamic topology, wireless channel characteristics, mobility, 978-0-7695-4682-7/12 $26. 0  © 2012 IEEE DOI 10. 1109/UKSim. 2012. 87 569 density of the nodes etc. ), analytical methods may become inappropriate as they require certain simpli? cations to model and predict the performance of the system. The simpli? cations may lead to inaccurate results in case of unrealistic assumptions [2],[3]. Experimental studies such as [4],[5],[6],[7] are not always practical for evaluation of systems with different architectures and under various conditions, mainly because of the dif? culties in deployment of real systems. Potential dif? ulties associated may be deploying tens or hundreds of sensor nodes in the physical environment, program the nodes and monitor their behaviour, the high costs involved in obtaining the instrumentation and other aspects such as fault tolerance, and scalability. It is well known that when it comes to benchmarking, the results in many cases cannot be extrapolated to suit the changes in the system or environment. Hence, testing and performance evaluation of WSNs through analytical modelling, real deployment and test beds can become complex, inaccurate, time consuming and/or costly.Simulation is currently the most widely adopted method for analyzing WSNs. Simulation studies provide quicker evaluation, optimisation and modi? cations of the proposed algorithms and protocols at design, development and implementation stages. A number of simulation tools are available with different features, models, architectures and characteristics for performance evaluation in WSNs. Packet level simulators offer various optimisation methods for free space scenarios and avoid the effects of path loss 1 that may be caused by different obstacles. The existing studies considering path loss for WSNs on the other hand are quite conservative.The impacts of path losses are not considered, and analysed together with details in upper layers such as network and application. In this paper, a new approach is considered to combine 1 Path loss is the attenuation in power density of an electromagnetic wave as it propagates through space. the path loss related issues with packet level simulation. A case study is presented which uses path-loss as well as network and application layer data in order to predict the network lifetime. Well known path loss computation models are adopted to use with a new tool, which allows the users to deploy ensors in a two dimensional abstraction of the physical environment, and to introduce obstacles. The new tool in turn communicates with well-known Castalia package and OMNET simulation environment. The energy consumption of the nodes considering the impact of path loss for different transmission powers is presented, the tradeoff between traditional performance measures such as packet loss and residual energy is illustrated. The approach presented lends itself as a ? exible and ef? cient tool which provides a more realistic approach for analysing WSNs and evaluating the performance in terms of energy ef? iency. The ? exibility of abstraction provided for the physical environment, also makes it possible to use various path loss models (even experimental ones). The rest of the paper is organised as follows: Section II considers various types of simulators. In section III, our approach is presented. Section IV provides the details of home automation application which is chosen as a case study. Section V shows the numerical results obtained. The impact of path loss on energy consumption of the nodes in a WSN is shown as well as the behaviour of nodes for different transmission powers in presence of path losses.In section VI, conclusion and future studies are presented. II. R ELATED W ORK In this section, we consider existing simulators. They can be classi? ed based on their level of complexity in to three main categories: Instruction, algorithm and packet level. A. Instruction level simulators Instruction level simulators are often regarded as emulators. They model the CPU execution at the level of instructions. TOSSIM [8], Atemu [9], Avrora [10] are well known emulators. TOSSIM is the most commonly used emulator. However, compared to other emulators, it is not the most precise one. TOSSIM, is a platform speci? simulator (a TinyOS mote simulator) which can compile any code written for TinyOS to an executable ? le. TinyViz, is the basic GUI for TOSSIM which can visualize and interact with the running simulations. TOSSIM is speci? c for TinyOS applications on Mica motes sensors and do not include power models. Avora, is a javabased emulator used for programs speci? cally written for AVR microcontrollers produced by Amtel and the Mica2 sensor modes. Atemu provides low-level emulation of the operation of individual sensor nodes. A unique feature of Atemu is its ability to simulate a heterogeneous sensor network.It is scalable and its high ? delity platform is used as a predeployment tool for sensor networks. B. Algorithm level simulators Shawn [11], AlgoSensim [12], and Sinalgo [13], are well known algorithm level simulators with emphasis on the logic, data structure and presentation of the algorithms. They rely on some form of graphical data structure to demonstrate the communication between the nodes. Shawn is a very powerful tool in simulating large scale networks with an abstract point of view. It supports distributed protocols and generic high level algorithms. AlgoSensim focuses on network speci? analysis of algorithms like localization, distributed routing, and ? ooding. AlgoSensim mainly facilitates the implementation and quality analysis of new algorithms. Sinalgo focuses on the veri? cation of network algorithms and abstracts from the underlying layers. It also offers a message passing view of the network. Sinalgo can be employed for quick prototyping and veri? cation in freely customizable network settings. C. Packet level simulators OPNET, Qualnet, NS-2, GloMoSim, are some of the most commonly used packet level simulators. They implement the data link and physical layers in the OSI network layers.Hence, radio models, 802. 11b or newer MAC protocols, fading, collisions, noise and wave diffractions are commonly implemented. Network Simulator (NS) is a discrete event simulator written in combination of C++ and OTcl. OTcl is an object oriented scripting language, developed mainly for networking research. It provides extensive support for simulation of TCP, multicast protocols, and routing for wired and wireless networks. With protocol implementations being widely produced and developed, the extensibility of NS-2 has been a major contributor to its success.It has an object-oriented design which allows for easy creation of new protocols. The key features for WSNs include battery models, hybrid simulation support, sensor channels, scenario generation tools and a visualization tool [14]. Scalability, lack of application model and the lack of customization are few limitations of NS-2 along with lacking an application model [3]. OPNET [15] and Qualnet [16] are commercialized network simulator software with powerful standard modules and they provide good simulation environment.OPNET is an excellent choice to simulate Zigbee based networks with the implementation of Zigbee protocol and IEEE 802. 15. 4 MAC protocol. Qualnet performs well in simulating large scale sensor networks due to its scalability in wireless simulation, but OPNET simulation requires a long time when the number of sensors considered is large. The above mentioned simulators use rather simple radio/channel models [17]. Also, the simulators are still platform speci? c and moderately scalable, making them unsuitable for protocol /algorithm design and testing.Furthermore the environmental details and especially the effects of path loss has not been considered in any of the given simulation packages. III. O UR APPROACH Figure 2 outlines the main components of our approach. This has been implemented in a tool called PlaceLife. An environment editor allows the user to specify the physical environment by using a graphical editor. The environment can include different obstacles and different sensors. An obstacle can have different properties such as the material it is 570 PlaceLife other layers info Environment Editor Application Model Path loss Model Translation engine ommonly used path loss models that de? nes the behaviour of signal strength in an indoor area. The path loss behaviour is dependent on the distance between nodes and the attenuation factor added by the objects. The attenuation can vary based on several factors such as the construction materials (e. g. , wood, glass and concrete) and the object size. In the Table I we show some attenuation values in dB introduced by various materials. We provide a detailed discussion in the next Section. The dependant path loss model can be expressed as [21]: LP = L0 + 20log(d) + mtype wtype Castalia Omnet++ Fig. 2. PlaceLife here, LP represents the path loss between two nodes, d is the distance between the two nodes, L0 is the path loss in free space environment, mtype refers to the number of objects of the same type and wtype is the loss in decibels attributed to that particular object. B. The translation engine The translation engine takes as an input the environment, application, and path loss models in order to produce simulation scripts. We use Castalia [22] as a simulation tool. Castalia is a WSN simulator used for initial testing of protocols and/or algorithms with a realistic node behaviour, wireless channel and radio models.Since Castalia is highly tunable and can simulate a wide range of platforms, it is used to evaluate different platform characteristics. Castalia features an accurate radio model based on the work of the authors in [23]. It also features physical process model, considering clock drift, sensor energy consumption, CPU energy consumption, sensor bias etc. Unpredictability of the wireless channel, energy spent in transmission/receiving packets, performance degradation experienced by duty cycles, collisions are usually overlooked by simple simulators.However these details are well established in Castalia [17]. All main components that affect the energy consumption of sensor nodes are considered that are the micro-processor, the sensor module, wireless transmitter/receiver and the path loss. We emphasise that while Castalia provides a good low level simulation platform; it does not provide any means to specify the application behaviour, the environment and the path loss models. The application behaviour is needed to derive application level simulation parameters. The environment and the path loss models allow the calculation of the path loss.In fact while Castalia assumes that the user provides path loss related parameters, our approach automatically derives those values from high level models such as the environment and path loss. IV. H OME AUTOMATION Monitoring and automatic control of building environment is a case study considered quite often [24], [25], [26], [27]. Home automation can include the following functionalities: (i) heating, ventilation, and air conditioning (HVAC) systems; (ii) emergency control systems (? re alarms); (iii) centralized control lighting; and (iv) other systems, to provide comfort, energy ef? ciency and security.In order to validate our approach made of and its size. The environment editor also allows the speci? cation of the sensor position in the physical environment. Obstacles and sensor position are used to compute the path loss. An application model de? nes the behaviour of nodes. From this model various performance parameters such as transmission and sensing rates can be derived. PlaceLife considers information from other layers such as network, data and physical layers to have a more realistic approximation for the life time. At network layer different protocols such as AODV [18] and DSR [19] can be speci? d but also static routing can be de? ned. This can be easily speci? ed on the environment model. Although various data link layer access methods can be used, the Timeout MAC (T-MAC) has been chosen in this case study. T-MAC is a contention based MAC protocol that use synchronised sleep schedules between the nodes in a WSN to conserve energy [20]. Also T-MAC provides both collision avoidance and reliable transmission. A. Path loss Path loss is the attenuation in power density of an electromagnetic wave as it propagates. Path loss is consequence of many effects such as free-space loss, refraction, diffraction, re? ction, aperture-medium coupling loss, and absorption. Path loss is also affected by other factors such as propagation medium (dry or moist air), the distance between the transmitter and the receiver, and the frequency of the signal. When the effects of path loss are not considered, the evaluation of underlying structure can become optimistic, since the problems associated, retransmissions and the way this phenomena affects the energy consumption are not taken into account. In our approach a path loss model can be speci? ed by the user. This model is used together with the physical environmental model in order to de? e the path loss between two nodes. In this paper we consider indoor environment and the dependant path loss model [21]. This is one of the most 571 A5 T Sm Sp Sp A3 BS A1 T T A4 Sp Sm Sm T Sp = sprinkler T = temperature Sm = smoke BS = base station concrete wood glass Sm Sp T A2 Sp Fig. 4. Fig. 3. Home automation Energy consumed by each node with and without path loss we consider the ? re alarm system and the automatic heating application. The ? re alarm system is composed of different temperature sensors and smoke detectors that are distributed inside the building.There are also sprinkler actuators used to enable the water ? ow in case of ? re. All the temperature sensors monitor the temperature at regular intervals (every 30 seconds). When a temperature sensor reads a value that exceeds a speci? ed threshold; it sends an alert message to the smoke detector. The smoke detector receives the alert and checks for smoke. An alarm is raised when the smoke is detected. In this case the smoke sensor also activates all the sprinklers. The automatic heating application is composed of different temperature sensors, a base station and various heaters.The temperature sensors send readings every 30 seconds to the base station. This is placed at the center forming a star topology. The base station averages the readings and decides whether or not the central heating system should be on. More speci? cally the base station works in the following way: if the heating is turned on and the average temperature is greater than the minimum threshold, the central heating system turns off. if the average temperature is less than the minimum threshold, the central heating system turns on. We consider the scenario of Figure 3. A ? at composed of ? e rooms (A1-A5). We also consider different obstacles such as wooden doors, walls and glass partition. V. N UMERICAL RESULTS AND DISCUSSIONS In order to show the usefulness and effectiveness of our approach and to analyse various factors affecting the performance in terms of energy consumption of WSNs, the numerical results are presented in this section. The simulation parameters are as follows: CC2420 radio de? ned by the Texas instruments is used, the output power of the different transmission levels in dBm are varied from 0 to -25dBm. Energy consumption for each transmission level varies.For instance for 0 dBm power consumed for listening (receiving) is 62 mW and for transmission is 57. 42 mW. Packet rate is kept at 250 kbps, the radio bandwidth is 20 MHz and the simulation runs for 9000 sec. T-MAC is used as a MAC protocol, and this makes the length of each frame period for all nodes 610 milliseconds, and the duration of listen time out 61 milliseconds. For our case study, we calculated the path-loss due to the material and explicitly set our path loss map [21], [28]. Refer to Figure 3 and Table I [21] for each type of obstacle and for its contribution to path loss.For the sake of the presentation we use numbers to represent sensors. Node 0 represents the base station. Nodes 1,4,5,7, and 9 monitor temperature in areas A1,A5,A4,A3, and A2 respectively. Nodes 2,3,6, and 8 monitor smoke in the areas A1,A5,A4, and A3 respectively. Table II and Table III show the energy consumed by the nodes for the application scenario considering the path-loss phenomenon and ignoring the path loss respectively. Similarly, Figure 4 shows the difference in energy consumed by each node for two different cases. In case one path loss is ignored, and for the next set of results the path loss is present.It is evident that the lifetime of the nodes is heavily TABLE I PARTITION DEPENDENT LOSSES FOR 2. 4 G HZ obstacles Concrete wall Wooden door Glass wall Cinder wall window Brick Masonry brick metal door attenuation in dB 12 2. 8 2 4 2 5 17 12. 4 TABLE II E NERGY CONSUMED BY THE NODES IN JOULES , CONSIDERING PATH LOSS nodes 0 1 2 3 4 5 6 7 8 9 energy 100. 7 84. 9 95. 6 94. 3 90. 1 88. 8 89. 3 88. 9 90. 5 91. 2 TABLE III E NERGY CONSUMED BY THE NODES IN JOULES , IGNORING PATH LOSS nodes 0 energy 81. 4 1 2 81. 4 82. 6 3 4 5 6 7 8 9 81. 4 81. 5 81. 5 82. 7 81. 4 82. 4 83. 1 572 Fig. 5. Energy consumed vs. ransmitted power for nodes 5-9 Fig. 6. Energy consumed vs. transmitted power vs. packets lost dependent on the impact of the path loss, and ignoring the effect of path loss would be an optimistic assumption when energy consumed by each node is considered. This is because, when the effects of path loss are not considered, problems associated, retransmissions and the way this phenomena affects the energy consumption are not taken into account. However these factors affect the life time of the node. Node 3 consumes 13 joules of more energy due to path loss, when compared to no path loss.Figure 5 shows the life time of the nodes 5 to 9, considering the impact of path loss for different transmission powers. Transmission power is varied from -25 dBm to 0 dBm, the energy consumption of the nodes is increased as we increase the transmission power. For node 7, as the transmission power is increased from -25 dBm to 0dBm, the energy consumed by the node also increases from 80. 1 joules to 88. 9 joules. The trade-off between traditional performance measures such as packet loss and residual energy is presented in Figure 6.The dotted lines represent the packets lost and the straight lines represent the energy consumed by each node. As the transmission power is decreased from 0 dBm to -25 dBm, there is a gradual increase in amount of packets lost. For node 0, as the transmission power is decreased from 0 dBm to -25 dBm, the number of packets lost increases to 370, from 206 and the energy consumed increases to 100 joules, from 88 joules. Because of the retransmissions, more energy is consumed by the nodes. But the increase in transmission power does not necessarily mean increase in the life time as there are no retransmissions.When the tradeoff between the packet loss and the energy consumed is analysed, it can be seen that the optimum transmission power should be between -15 to -5 dBm where the energy consumption is less than 95 joules and packet loss is less than 200 packets. VI. C ONCLUSION AND FUTURE WORK In this paper, a simulation-based study is presented that uses path-loss network and application layer data in order to predict the network lifetime. Physical environment is considered as well. We show that when path-loss is introduced, increasing the transmission power is needed to reduce the amount of packets lost.This presents a tradeoff between the residual energy and the successful transmission rate when more realistic settings are employed for simulation. It is a challenging task to optimise the transmission power of WSNs, in presence of path loss, because although increasing the transmission power reduces the residual energy, it also reduces the number of retransmissions required. This work is by no means complete. The concept is far more complicated than just ? nding out the network life time. The main focus is to place the nodes in a way to maximise the network life time, which is the major constrain of WSNs. Work is in progress.R EFERENCES [1] I. Akyildiz, S. Weilian, Y. Sankarasubramaniam, and E. Cayirci, A survey on sensor networks, Communications Magazine, IEEE, vol. 40, no. 8, pp. 102 – 114, aug 2002. [2] T. Krop, M. Bredel, M. Hollick, and R. Steinmetz, Jist/mobnet: combined simulation, emulation, and real-world testbed for ad hoc networks, in Proceedings of the second ACM international workshop on Wireless network testbeds, experimental evaluation and characterization, ser. WinTECH ’07. New York, NY, USA: ACM, 2007, pp. 27–34. [Online]. Available: http://doi. acm. org/10. 1145/1287767. 1287774 [3] G. Chen, J. Branch, M. J. P? g, L. Zhu, and B. K. Szymanski, Sense: A wireless sensor network simulator, 2012. [4] K. Phaebua, T. Lertwiriyaprapa, C. Phongcharoenpanich, and M. Krairiksh, Path loss prediction in durian orchard using uniform geometrical theory of diffraction, in Antennas and Propagation Society International Symposium, 2009. APSURSI ’09. IEEE, june 2009, pp. 1 –4. [5] M. -S. Pan, L. -W. Yeh, Y. -A. Chen, Y. -H. Lin, and Y. -C. Tseng, A wsn-based intelligent light control system considering user activities and pro? les, Sensors Journal, IEEE, vol. 8, no. 10, pp. 1710 –1721, oct. 2008. [6] M. Halgamuge, T. -K. Chan, and P.Mendis, Experiences of deploying an indoor building sensor network, in Third International Conference on Sensor Technologies and Applications, 2009. SENSORCOMM ’09. , june 2009, pp. 378 –381. [7] S. Shuo, S. Hao, and S. Yang, Design of an experimental indoor position system based on rssi, in 2010 2nd International Conference on Information Science and Engineering (ICISE), dec. 2010, pp. 1989 –1992. [8] P. Levis, N. Lee, M. Welsh, and D. Culler, Tossim: accurate and scalable simulation of entire tinyos applications, in Proceedings of the 1st international conference on Embedded networked sensor systems, ser.SenSys ’03. New York, NY, USA: ACM, 2003, pp. 126–137. [Online]. Available: http://doi. acm. org/10. 1145/958491. 958506 [9] J. Polley, D. Blazakis, J. McGee, D. Rusk, and J. Baras, Atemu: a ? ne-grained sensor network simulator, in First Annual IEEE Communications Society Conference on Sensor and Ad Hoc Communications and Networks, 2004. IEEE SECON 2004. , oct. 2004, pp. 145 – 152. 573 [10] B. L. Titzer and et al. , Avrora: Scalable sensor network simulation with precise timing, in IN PROC. OF THE 4TH INTL. CONF. ON INFORMATION PROCESSING IN SENSOR NETWORKS (IPSN, 2005, pp. 477–482. [11] (2012, Jan. ) Shawnwiki. Online]. Available: http://shawnwiki. coalesenses. com/index. php/Shawn Introduction [12] Jacques and Marculescu. (2011, Dec. ) Algosensim. [Online]. Available: http://tcs. unige. ch/doku. php/code/algosensim/overview [13] (2011, Dec. ) Sinalgo. [Online]. Available: http://dcg. ethz. ch/projects/sinalgo [14] T. Issariyakul and E. Hossain, Introduction to Network Simulator NS2, 1st ed. Springer Publishing Company, Incorporated, 2008. [15] (2011, Dec. ) Opnet. [Online]. Available: opnet. com [16] (2011, Dec. ) Qualnet. [Online]. Available: scalablenetworks. com [17] D. Kotz, C. Newport, R.S. Gray, J. Liu, Y. Yuan, and C. Elliott, Experimental evaluation of wireless simulation assumptions, in MSWiM ’04: Proceedings of the 7th ACM international symposium on Modeling, analysis and simulation of wireless and mobile systems. New York, NY, USA: ACM, 2004, pp. 78–82. [Online]. Available: http://dx. doi. org/10. 1145/1023663. 1023679 [18] Z. Che-Aron, W. Al-Khateeb, and F. Anwar, The enhanced faulttolerant aodv routing protocol for wireless sensor network, in 2010 Second International Conference on Computer Research and Development, may 2010, pp. 105 –109. [19] H. Ben Chikha, A. Makhlouf, and W.Ghazel, Performance analysis of aodv and dsr routing protocols for ieee 802. 15. 4/zigbee, in International Conference on Communications, Computing and Control Applications (CCCA), 2011, march 2011, pp. 1 –5. [20] W. Ye, J. Heidemann, and D. Estrin, Medium access control with coordinated adaptive sleeping for wireless sensor networks, IEEE/ACM Transactions on Networking, vol. 12, no. 3, pp. 493 – 506, june 2004. [21] K. Pahlavan and P. Krishnamurthy, Networking Fundamentals. Chichester, UK: John Wiley and Sons, 2009. [22] (2011, Dec. ) Castalia. [Online]. Available: http://castalia. npc. nicta. com. au [23] M.Zuniga and B. Krishnamachari, Analyzing the transitional region in low power wireless links, in First Annual IEEE Communications Society Conference on Sensor and Ad Hoc Communications and Networks, 2004. IEEE SECON 2004. , oct. 2004, pp. 517 – 526. [24] D. -M. Han and J. -H. Lim, Smart home energy management system using ieee 802. 15. 4 and zigbee, IEEE Transactions on Consumer Electronics, vol. 56, no. 3, pp. 1403 –1410, aug. 2010. [25] K. Gill, S. -H. Yang, F. Yao, and X. Lu, A zigbee-based home automation system, IEEE Transactions on Consumer Electronics, vol. 55, no. 2, pp. 422 –430, may 2009. [26] Y.Tachwali, H. Refai, and J. Fagan, Minimizing hvac energy consumption using a wireless sensor network, in Industrial Electronics Society, 2007. IECON 2007. 33rd Annual Conference of the IEEE, nov. 2007, pp. 439 –444. [27] H. S. Kim, J. -H. Song, and S. Lee, Energy-ef? cient traf? c scheduling in ieee 802. 15. 4 for home automation networks, IEEE Transactions on Consumer Electronics, vol. 53, no. 2, pp. 369 –374, may 2007. [28] C. Anderson and T. Rappaport, In-building wideband partition loss measurements at 2. 5 and 60 ghz, IEEE Transactions on Wireless Communications, vol. 3, no. 3, pp. 922 – 928, may 2004. 574