International Journal of Computer Information Systems, Vol. 3, No. 2, 2011 Comparative Study and Design of a Multihop Protocol in a Wireless Body Area Network towards Energy Efficiency 1 D.V.Srihari, 2 T.Ravi Kumar, 3 V.Karunakar Reddy, 4 K.Naresh 1 Assoc. Professor Dept. of ECE, SKTRM College of Engineering, Kondair, MBNR (Dt.), A.P, Student, M.Tech(ECE), Dept. of ECE,SKTRM College of Engineering, Kondair, MBNR (Dt.), A.P, 3 Student, M.Tech(ECE), Dept. of ECE, SKTRM College of Engineering, Kondair, MBNR (Dt.), A.P, 4 Student, M.Tech(ECE), Dept. of ECE,SKTRM College of Engineering, Kondair, MBNR (Dt.), A.P. 2 Abstract : - Wireless Body Area Networks (WBANs) have characteristic properties that should be considered for designing proper network architecture. Low quality wireless links, and demand for a robust data transmission scheme at low energy are important issues in WBANs. Using ultra low power wireless transceivers to reduce power consumption causes a limited transmission range. This implies that a multihop protocol is a better design choice. In this paper we design and compare a basic multihop scheme against a single-hop protocol stack in a duty cycling MAC and other conditions to analyse the effect such changes have on the throughput. It is our goals here to improve the throughput by using lesser transmit power and a better routing mechanism. The discrete simulation framework Omnet++ was used to design the algorithm and conduct the experiments. Keywords : Wireless Body Area Network, Omnet++. I. INTRODUCTION A Wireless Body Area Network (WBAN) consists of tiny sensor nodes that measure the vital signals of the body, such as blood pressure, fever, heart beat rate, and movement activities. The information is relayed to a gateway node on the body and then to a base station for further analysis. Important issues to be considered while designing a network architecture and protocol stack for this kind of wireless networks are as follows – 1. Power constraint and short RF transmission range: The power and size constraints of body sensors are tighter than in other wireless sensor applications. A short transmission range is a common limitation of low-power RF transceivers. Therefore the amount of consumable energy is strictly limited. 2. Low quality of wireless links: Several investigations using simulations and experiments show that the propagation loss around and in a human body is considerably high. 3. The general conclusion is that wireless links in WBANs are more unreliable than in other Wireless Sensor Networks. August Issue II. PROTOCOL STACK DESIGN FOR WBAN The challenges in WBANs demand particular attention while designing an appropriate protocol stack. On the other hand, the limited network size provides opportunities to relax some constraints on the protocol. To realize an appropriate protocol stack, we use a multihop ring based mechanism for data routing on top of a duty-cycled MAC layer. Assume that there are 6 sensor nodes deployed on a body each with a unique ID from 0 to 5.All these nodes are of the same type. The nodes are #0 at right hip , #1 at left wrist , #2 at right wrist,#3 at left ankle,#4 at right ankle, #5 at chest. A path loss map is constructed based on empirical measurements on a WBAN. The protocol stack design follows a layered model. The application layer consists of a “throughputTest” model where all nodes send packets to a sink/hub node at a constant (configurable) rate. The hub is node 0. The network layer consists of a multihop mechanism for data transmission. The MAC layer implements a simple CSMA mechanism where duty cycle and some other parameters can be controlled. III. DATA ROUTING USING A MULTIHOP APPROACH In this algorithm, nodes do not have a specific parent. A node just gets a level number (or ring number) during setup. The first setup packet sent from the sink has level 0.Any node that receives it adds 1 to the level and retransmits it. The process continues with every node adding 1 to the level of the received packet. Eventually all connected nodes will have a level number (there is also a possibility for unconnected nodes). When a node wants to send a packet to the sink it does not send it to a particular node but rather broadcasts it, attaching its level number. Any node with a smaller level number will rebroadcast it. The process continues until the sink is reached. Therefore the creation of multiple paths to the sink node provides robustness to the routing strategy. In comparison to single route algorithms this is quite reliable but if traffic Page 116 of 130 ISSN 2229 5208 International Journal of Computer Information Systems, Vol. 3, No. 2, 2011 passes a certain threshold, congestion will affect the performance. Table 1 shows the configuration for the WBAN. In addition TX Power levels, a delay transition matrix, and a power transition matrix are also defined. As seen with -10dBm maximum throughput is achieved. Range is normally in (93, 99) percent. With -12dBm throughput range is (90,100) percent .With -15dBM throughput range is (83.5, 99) percent .With -20dBM throughput range is (70, 98.7) percent One can observe that between nodes 0 & 2, throughput is very high in all cases i.e. between (R-hip & R-wrist). Between 0 and 4, throughput on average is 234(R-hip & R-ankle) and between 0 and 1, throughput on average is 229.5(R-hip & Lwrist). Therefore these are the links with best quality throughput in the system. Setting up paths using the above 3 links may therefore decrease the overall energy consumption of the network. In comparison to 0-4 link, the 0-1 link falls drastically when power is reduced to -20dBm.This fall may be due to the distance between the 2 nodes. The energy consumption varies little at 0.158 Joules. In 50 seconds each node sends 249 packets to node-0.So total number of packets sent to node-0 is 249*5=1245 packets. On an average, as TxOutputPower decreases average throughput varies between (205,242).So at -10dBm throughput is 97% whereas at -20dBm throughput has fallen to 82.3%.The objective would therefore be decreasing this range by better mechanisms. The average energy consumption is therefore 0.158 joules for the network. The total energy consumption is 0.158*6=0.948 Joules in 51 seconds. The table shows that the energy consumed by all nodes is uniform. The receiver node-0 listens all the time so it consumes 3.1mW * 51sec =0.1581 Joules. The sender nodes draw either 2.93mW (when transmitting) or 3.1mW (when just listening), to find the energy we need to know what percentage of the total time are they transmitting. We see that each sender node sends 5 packets per second. And the default size for packets = 130bytes = 1040 bits. At 1024Kbps data rate, a packet transmission takes approximately 1ms. So only 5msec out of a second is a sender node in TX mode. This means that the total energy consumption of a sender node is: (0.005*2.93mW + 0.995*3.1mW) *51 sec = 0.15805 Joules The total energy consumed for the network is given by [(Time*TrasmitPower+Time*TransitionPower+Time*IdleLi st ening_Power)*No. of Nodes]. August Issue Page 117 of 130 ISSN 2229 5208 International Journal of Computer Information Systems, Vol. 3, No. 2, 2011 The transition time is 20micro seconds per transition, so it is 0.2msec in a second where 5 transmissions happen. The transition power is 3mw (in between 2.93 and 3.1) so the effect at the end energy number is extremely less. It is important to note here that energy consumption in the network should not be skewed leading to premature death of some nodes with maximum load. Therefore energy consumption should be properly distributed across all nodes. Therefore it helps if there are nodes that periodically load balance the packet distribution process. As per simulations above , it can be seen that with dutycyle as 1.0, energy consumption always remains constant and it never decreases, instead the throughput gets affected due to interference and fading of packets. Therefore the key to minimizing energy consumption of a node is by tuning the duty cycle and thereby minimizing idle listening. The next few experiments show the results of the various duty cycles:- The above table shows that on an average the energy consumption per node has decreased after duty cycling the MAC. Therefore while designing the protocol stack our main objectives would be – 1) Building a packet dissemination mechanism that can yield a high throughput than the baseline above, for a duty cycle of 1.0 and with the minimum TxOutputPower i.e. -20dBm. 2) The next step would be to devise a strategy that does not affect the throughput much yet can work with the least duty cycle of 0.2. By our multi-hop mechanism, we now improve the throughput as below – As seen with duty cycle as 1.0, the results are the same as earlier cases. With anything below 0.8, the throughput drops extremely low. The best among these values is 132.8 at 0.6 duty cycle and -10dBm power. On duty cycling the MAC, beacon packets are sent before data. This increases the packet flow in the network increasing chances of collision. Also as the duty cycle is decreased the beacons increase. So the single hop throughput range of (205,242) was now improved by having an 86.3% improvement as compared to the earlier 82%. Also more robustness is now achieved since at all power levels throughput has increased by a multi-hop approach. This is because multiple paths to the sink are now possible. The average energy consumption still remained as 0.158 Joules since the experiment was done with a duty cycle of 1.0. August Issue Page 118 of 130 ISSN 2229 5208 International Journal of Computer Information Systems, Vol. 3, No. 2, 2011 Therefore with a multi-hop approach the robustness and reliability was dealt with at the routing layer with the same energy consumption as the single hop approach. V. CONCLUSIONS This paper has presented a basic study and design factors to be taken into consideration towards energy efficiency improvement for a WBAN. While designing the protocol stack our main objectives were as follows – • To build a packet dissemination mechanism that can yield a high throughput than the baseline above, for a duty cycle of 1.0 and with the minimum TxOutputPower i.e. - 20dBm.This was achieved by the multihop protocol as described earlier. • As future work, the next step would be to devise a strategy that does not affect the throughput much, yet can work with the least duty cycle of 0.2.For this, adjustments towards the MAC layer are still required. 2]. T.Ravi Kumar, Student M.Tech(ECE), SKTRM College of Engg & Tech., Dept .of ECE, Kondair, A.P. His research interests in areas of Computer Networks, Mobile Computing, DSP, MATLAB, Data Mining, Sensor Networks. He attend so many Workshops and National & International Conferences 3]. V.Karunakar Reddy, Student M.Tech(ECE), SKTRM College of Engg & Tech., Dept .of ECE, Kondair, A.P. His research interests in areas of Computer Networks, Mobile Computing, DSP, MATLAB, Data Mining, Sensor Networks. He attend so many Workshops and National & International Conferences REFERENCES: [1] OMNeT++ website. http://www.omnetpp.org. [2] B. Braem et al. The wireless autonomous spanning tree protocol for multihop wireless body area networks. In Proc. of 3rd Int’l Conf. on Mobile and Ubiquitous Systems (MobiQuitous), pages 284–291. IEEE, July 2006. [3]E. Jovanov et al. A wireless body area network of intelligent motion sensors for computer assisted physical rehabilitation. In Journal of NeuroEngineering and Rehabilitation, pages 2–6, March 2005. . atr et al. A low-delay protocol for multihop wireless body area networks. In Proc. 4th MobiQuitous, pages 1–8. IEEE, 2007. 5 David . Johnson and David A.Maltz, “Dynamic source routing in Ad hoc wireless networks”. In Mobile Computing, edited by Tomasz Imielinski and Hank Korth, chapter 5, pages 153-181. Kluwer Academic Publishers, 1996. [6] A. Natarajan et al. Investigating network architectures for body sensor networks. In Proc. of the 1st ACM SIGMOBILE Int’l workshop on Systems and networking support for healthcare and assisted living environments, pages 19–24. ACM, 2007. 4].K.Naresh, Student M.Tech(ECE), SKTRM College of Engg & Tech., Dept .of ECE, Kondair, A.P. His research interests in areas of Computer Networks, Mobile Computing, DSP, MATLAB, Data Mining, Sensor Networks. He attend so many Workshops and National & International Conferences. AUTHORS PROFILE: 1].D.V.Srihari, M.Tech(ECE), Assoc. Professor, SKTRM College of Engg & Tech., Dept .of ECE, Kondair, A.P. His research interests in areas of Computer Networks, Mobile Computing, DSP, MATLAB, Data Mining, Sensor Networks. He attend so many Workshops and National & International Conferences. August Issue Page 119 of 130 ISSN 2229 5208