INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004

1

A Grid-Based Location Estimation Scheme using Hop Counts for Multi-hop Wireless Sensor Networks Joo Ghee Lim, S V Rao, Senior Member, IEEE Institute for Infocomm Research Singapore 117674 Abstract— This paper describes a grid-based location

estimation scheme for multi-hop wireless sensor networks (GRIPHON) using the hop count information of the sensor nodes. The scheme makes use of the correlation between a node’s position in the network region and the number of hops that node is away from a fixed number of reference nodes (markers) located within the region, known as a hop count vector. A node is predicted to be located in a grid zone with a hop count vector value that corresponds with the hop count vector of the node. Simulation results show that for a small number of markers, GRIPHON is able to predict accurately the location of nodes and the scheme scales well in situations of increased node density and area. We also describe an experimental setup used to test our scheme. Index Terms— location estimation, sensor networks

R

I. INTRODUCTION

ecent advances in MEMS (MicroElectroMechanical Systems) technology, communication designs over RF and optical media as well as low power management have spurred the development of relatively low-cost, low-powered wireless sensors. These sensors can effectively be deployed over an area by the hundreds or thousands. Within the area, the sensors can communicate with each other, forming a network. There has been much research into the many aspects of these wireless sensor networks, e.g. power-aware MAC protocols [1,2], energy-conserving routing schemes [3,4] and other sensor-related communication protocols [5,6]. In many of these applications, the location of the sensor node is needed before the information gathered can make sense. In fact, some routing schemes, e.g. proposed in [6], make use of the sensor node’s location to increase routing efficiency. Using GPS-enabled location scheme may not be feasible due to cost, processing power and locality (indoor) constraints. This paper introduces GRIPHON – GRId-based Positioning using HOp couNts. GRIPHON seeks to predict the location of sensor nodes with respect to rectangular grid positions within the network area by using the hop counts these nodes are away from a number of fixed, known markers.

Section II of this paper discusses some of the related works in current location estimation research and section III gives an overview of GRIPHON. Some simulation results will be presented in section IV. In section V, we give a brief description of the experimental setup that we used to test our scheme. II. RELATED WORK Current location estimation techniques can generally be divided into two main categories – Direct and Inferred. In the Direct approach to location estimation, there is usually a finite set of reference nodes or base stations, with known locations, that all other nodes (with unknown locations) communicate directly with to find their location. An example is the 3D-iD system described in [7], which uses GPS-like techniques to locate the RFID tags. Other similar approaches can be found in [8,9]. One shortcoming of the direct location estimation approach is the balance between the number of reference nodes and the signal coverage. In order to determine location, the coverage of the signal from the reference nodes must overlap. Using a smaller number of reference nodes will require longer transmission ranges, which will aggravate the multipath problems and increase measurement inaccuracies. However, having reference nodes with smaller coverage will require a higher density of reference nodes to be deployed, leading to scalability and cost issues. The main idea of the Inferred approach is to use the known, recently estimated locations (or at least some metric) of some sensor nodes to determine or infer the location of their neighbors (i.e. nodes that are connected to them). A node will receive ranging and location information from its neighbors to find its location. The results will then be transmitted to its own neighbors for them to perform the same operation, as is discussed in [10,11] The merit of the inferred approach over the direct one is that few or no reference nodes are needed. However, the assumption is that the position of the solid node has been determined with a fair amount of accuracy. Otherwise, the precision of the location estimation is seen to deteriorate over time and space. III.

SCHEME OVERVIEW

In this section, we describe the GRIPHON scheme for location estimation in a multi-hop wireless sensor network. As

INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004 discussed in section II, a direct location estimation approach requires base stations or beacons having long range, leading to scalability and multipath issues; whereas the inferred approach may cause errors to propagate and increase indefinitely. In GRIPHON, we hope to reduce these shortcomings. Essentially, GRIPHON uses some reference nodes, which we call markers placed within the network region. These markers, with fixed and known locations, seek to give the sensor nodes references from which location information can be computed. However, unlike the direct approach, the reference nodes are exactly like the sensor nodes except that they are fixed and contain known location information. Information propagates from the markers to the nodes throughout the network as each node communicates with its neighbors via the Smallest Hop Count Discovery algorithm described below. A. Smallest Hop Count Discovery The Smallest Hop Count Discovery scheme is the process where each node attempts to discover the smallest hop count it is away from each of the markers (i.e., its hop count vector). The algorithm works as follows: 1. Marker Mi, will broadcast to its neighbors, informing that it is a marker. 2. Each node, when it receives the hop count information, will increase this value by 1 hop count and compare it with what it currently contains (if it is available). If it is smaller, the information will be updated and forwarded to its neighbors. Otherwise, it will simply be discarded. 3. Eventually, every node will have information about the smallest hop counts it is away from each marker. This is represented by a hop count vector (hcv), HK = (h1, h2, h3, …, hi) where HK is the hcv of node K and hi is the smallest hop count from node K to marker i. The use of hop counts for location estimation is not new. Niculescu and Nath suggest this as one of the ways to calculate the distance a node is away from known landmarks in their Ad Hoc Positioning System [12]. Hop counts have also been extensively used in many routing protocols. In GRIPHON, we use the set of hop counts as a hcv to determine the correlation between the vector and the position of a node. B. Grid-based Positioning Through simulations, we discovered that there exists a correlation between the hcv and the position of the node within the network region - each small part of the network region can be uniquely identified with a range of hcv values. Using this property, we are able to predict the position of a node based on its hop count vector by comparing it with a standard hcv set for a particular network region of nodes (within an acceptable range of node density). The grid-based positioning scheme attempts to place a node into one of the smaller grids within a network region by referring to its hcv. This can be best explained by way of an example.

2

Fig. 1. The example network region

Fig. 1 shows the network region. It is typically a square region with markers at four corners. We assume that there are sensor nodes throughout this region. There is also a centralized reader that does the computation of the locations. We assume continuity in the network; i.e. all nodes have at least a direct or indirect path to the markers and centralized reader. The region is further subdivided into small grid zones of 5x5 m2. The positioning scheme works as follows: 1) Setup Phase In the setup phase, a standard set of mean-hcv for each grid zone is stored into the centralized reader. Where possible, the values of this mean-hcv set could be by deploying test nodes within the area and collecting the hcv information at each grid zone. Alternatively, the mean-hcv could be determined from simulations or modeling using similar node density and environment conditions (e.g., size and shape of network region). 2) Location Discovery Phase After the setup phase, sensors can be placed within the region and their locations determined as follows: After network connection is established (i.e., each node has communication links with its neighbors), smallest hop count discovery is performed to determine the hcvs of the nodes. These values are passed to the reader by way of some routing scheme. At the reader, the hcv of each node is compared with the mean hcv table to determine the grid zone the node is in. This is known as a threshold differencing operation. E.g., we compare a node, Ni with a hcv of (h1, h2, h3, h4)i with the mean hcv of (hM1, hM2, hM3, hM4)G where G is the grid zone. Ni is considered to be located in the grid zone x,y if h M 1 − h1 ≤ 1.0 hca; and h M 2 − h2 ≤ 1.0 hca;

and

h M 3 − h3 ≤ 1.0 hca;

and

h M 4 − h4 ≤ 1.0 hca.

1.0 hca is known as the threshold difference between the actual hcv and the mean hcv of nodes in the grid zone. A grid zone is considered to be a possible position of a node if the difference between the corresponding elements of the hcvs is not more than 1.0 hca. Varying this threshold difference value serves to tighten or loosen the criterion of position determination.

INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004

3

IV. SIMULATION RESULTS In this section, we present the results of simulations that have been done to analyze the performance of the GRIPHON scheme. The simulation has been done on a discrete event driven simulator written in C. A. Simulation Model In our simulation, we assume a circular radio propagation model that is similar for all the nodes. Depending on the radio range R, we create a circular region around a node of radius R. Nodes found within the region will have communication links with the node. Each node will thus be able to maintain a neighborhood set, consisting of nodes falling within the communication range. To simulate the link layer operation, we assume that there is a MAC protocol in place that prevents collision of messages sent by neighboring nodes. The simulation topology consists of a square region. One marker, mi with known position vector (xi,yi) is placed at each corner of the square region. The region is further divided into 5x5 m2 grid zones. Nodes will be placed anywhere within this region. Each node created consists of an actual location vector field, (xA, yA) indicating the actual position of the node. There is also a hcv field, (h1, h2, h3, h4) which is initially empty. For each simulation scenario, we first perform a setup phase where nodes are randomly placed within the region to achieve a similar node density. The nodes found in each grid zone are collected over a few simulation runs. The mean hcv is computed for 100 nodes in each of these grid zones to form the mean hcv table. In the actual simulation, nodes are placed randomly within the region and their hcvs found. Using the hcv table computed during the setup phase, we determine the position of each node with respect to the grid zones. To analyze the accuracy of the scheme, the predicted position is compared with the actual position (grid zone) of the node, calculated from the actual location vector. If the predicted position is found to be at most 1 grid zone away from the actual grid zone, we will consider it to be a right solution. From Fig. 2, assuming the predicted position (coordinate) of the node is in the center grid, the margin of error is found to be at most

(5 + 2.5)2 + (5 + 2.5)2

= 10.6 m away. This means that the

actual location of the node is at most 10.6 m away from the center of the predicted grid zone. With respect to the signal range, R of 5m, this gives us a confidence of locating the node approximately 2R away from its predicted location if it is a right solution. In our simulation, if the threshold differencing operation produced one and only one right solution, it is considered a correct prediction. If there are multiple solutions, we will also consider it a correct prediction as well if all the solutions

Fig. 2. Maximum error of location estimation (in m)

are right.a On the other hand, if there is at least one wrong solution, it is considered an incorrect prediction, since we are not confident of getting a solution that is within the 10.6 m error bound. If no solution can be found, we consider it to be a no-solution prediction. We compare the percentage of these different prediction outcomes from our simulations. B. Simulation Results We present here the results of two sets of simulations. In the first set, we vary the node density of the region while keeping the area constant at 50x50 m2. In the second set, we vary the area of the network region while keeping the average node density constant at 0.4 nodes/m2. In each simulation scenario, we collect the average results over 10 simulation runs. We found that the performance of the scheme is highly dependant on the value of the threshold difference. Setting too low a threshold difference (i.e. making the criterion stricter) has the effect of increasing the percentage of no-solution predictions, since we are not able to match to a suitable grid. Relaxing the threshold difference will decrease the occurrence of these no-solution cases. However, setting too high a threshold difference results in more multiple solution cases, where there may be chances where a wrong solution is found. There exists an optimal value where we balance the tradeoff between having actual solutions and not too many multiplesolution cases. We chose the value of 1.0 hca as the threshold difference to use in our simulations. 1) Vary Node Density In this simulation, we vary the node density of the network region to investigate the scalability of our scheme with respect to the number of nodes in a given region. This is important because most sensor network applications require typically hundreds to thousands of nodes within a region. Therefore, any location estimation scheme must be able to work under these conditions. Fig. 3 shows the percentage of correct, incorrect and nosolution predictions over different node density while keeping the area of the network region constant. In general, as the node density increases, the accuracy of the scheme’s prediction improves. This is the general characteristic of most a

Since all the solutions are right, we only need to choose any one of them to get a correct prediction.

INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004 location estimation scheme that makes use of inferred approach (i.e. where there is no direct measurement involved). At low node density, there is a greater chance when no

Fig. 3. Percentage of correct/incorrect/no-solution predictions for varying number of nodes in a 50x50 m2 area

Fig. 4. Percentage of nodes with different errors from actual location (varying node density)

solution is found because some nodes may require larger number of hops from the markers, causing a mismatch. The percentage of incorrect predictions remain fairly constant. Using the above simulation scenario, we perform another set of simulations to determine the extent of errors with regards to the absolute distance between the estimated and actual coordinates of each node. We assume that the predicted coordinate of the node is at the center of the grid zone. To eliminate multiple-solutions cases, one position is randomly chosen from these solutions. Where no solution can be found, the threshold difference of the highest hcv element is relaxed by 0.2; repeating the procedure with the next highest hcv element, until a solution is found. We vary the node density by repeating the simulation for 300, 500, 1000 and 1500 nodes within the same area of 50x50 m2. Fig. 4 shows the percentage of nodes with different absolute errors for a particular simulation scenario. It can be seen that, across all node densities shown, the absolute errors experienced by almost all the nodes do not exceed 10 meters (i.e. approximately 2R). It can also be seen that, the performance is constant over varying node density. This demonstrates the scalability of the scheme under increased node density. In fact, from the simulations, it was found that the performance degrades at low node densities (less than 300 nodes per 50x50 m2). The reason for this is because the hop count vector information becomes highly inconsistent as the

4

number of nodes within a given area decreases. Work is currently being done in modifying the scheme to work in these low node density situations.

Fig. 5 Percentage of correct/incorrect/no-solution predictions for different network region area with constant node density

Fig. 6. Percentage of nodes with different errors from actual location (varying area size)

2) Vary Area Size In the next set of simulations, we investigate the effect of network area size on the performance of our scheme by varying the area of the network region while keeping the average node density constant (at 0.4 node per m2). Especially in a big network area, we need to find out if the prediction errors will increase like in most inferred location estimation schemes. From Fig. 5, we can see that the accuracy of the prediction does not suffer as the area increase. This shows that our scheme does not suffer from the additive nature of errors found in most inferred location estimation approach. Moreover, the four markers (with a signal range that is similar to all the other sensor nodes) are apparently sufficient for accurate location prediction. The actual estimation error (in meters) experienced by each node, when we vary the network size, is also computed in another set of simulations. The multiple solutions and nosolution occurrences are eliminated as above. In the simulations, the errors experienced by the nodes are computed for area sizes of 30x30, 50x50, 60x60, 70x70 and 80x80 m2, while keeping the node density constant. Fig. 6 shows the percentage of nodes with different estimation errors for a particular simulation scenario. It can be seen that the performance of the location estimation scheme only degrades slightly as the area increases. Similarly, most of the nodes experience estimation errors of

INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004 less than 2R. V. EXPERIMENTAL IMPLEMENATION In this section, we give a brief description of the experimental setup that we have developed to test out our

irregularities of the nodes (e.g. multipath effects) and signal coverage non-uniformity caused by antenna–board interaction. Another challenge discovered is the fact the node’s signal boundary is not a definite cut-off point but a region where signal gradually grows weaker as one moves farther away

Fig. 8. Snap shot of the testbed implementation

location estimation scheme. We developed an implementation of our sensor nodes comprising of a RF transceiver chip with a microcontroller. The Chipcon CC1000 chipset was chosen for the transceiver because it has adjustable transmission power and channel frequency capability, as well as RSSI sensing ability. For the microcontroller, the Texas Instruments’ MSP430F149 was used as it provides for low power consumption, multiple power-saving modes and consists of a 16-bit architecture. These hardware features allow us to create a low-powered, flexible platform to build our schemes on. Above the hardware platform, we developed a MAC scheme, a routing scheme as well as the GRIPHON location estimation scheme. All the modules are interconnected with an event-driven scheduler as the operating system. Fig. 7 shows a version of the node we had created. Using the nodes, we developed a testbed of 40 nodes with identical response to test our schemes (GRIPHON, as well as the MAC and routing schemes). Each node is routed directly or indirectly to a centralized reader node that is connected to a notebook displaying the hcv values and translating these values into a position on the network region. Fig. 8 shows a snap shot of our testbed setup. Experiments conducted on the testbed are still ongoing but preliminary results show that we are able to estimate the positions of the nodes to within 1 grid zone away from its actual location 80% of the time. It may be noted that, from results of the testbed experimentation as well as simulation, the accuracy of GRIPHON is within 2R, where R is the signal range of the node. This level of accuracy is sufficient for most sensor network applications where information is required, not so much as from a specific position, but from an area of locality. For example, a user would be more interested in finding the temperature of the region occupied by a number of sensor nodes, as opposed to the temperature sensed by one of these nodes. Challenges faced when moving the scheme from a design to an actual implementation include signal propagation

5

Fig. 7. Snap shot of the node implementation

from the node. We are still studying the effects these physical issues have on GRIPHON specifically, as well as other position estimation schemes in general. Our testbed provides a useful platform to test and modify GRIPHON, along other sensor network protocols. [13] provides a more detailed account of the implementation of the testbed. VI.

CONCLUSION

This paper describes the use and implementation of a simple grid-based localization approach using hop counts. The simulations have shown that with a limited number of markers, the scheme is still able to scale well with increased area or node density. We also describe a testbed that we have developed to test our scheme. REFERENCES [1]

[2]

[3] [4]

[5]

[6]

[7] [8]

[9]

S. Singh and C.S. Raghavendra, “PAMAS – power aware multi-access protocol with signalling for ad hoc networks,” ACM Computer Communication Review, July 1998. K. Sohrabi and G.J. Pottie, “Performance of a novel self-organization protocol for wireless ad-hoc sensor networks,” in Proc. IEEE Vehicular Technology Conference, Vol. 2, 1999. J.H. Chang and L. Tassiulas, “Energy conserving routing in wireless adhoc networks,” in Proc. IEEE Infocom 2000, Vol. 1, 2000. I. Stojmenovic and X. Lin, “Power-aware localized routing in wireless networks,” in Proc. IEEE Int. Parallel and Distributed Processing Symposium, Cancun, Mexico, May 1-5, 2000. J. Kulik, W.R. Heinzelman and H. Balakrishnan, “Negotiation-based protocols for disseminating information in wireless sensor networks,” in Proc. 5th ACM/IEEE Mobicom Conference, Seattle, August 1999. W.R. Heinzelman, A. Chandrakasan and H. Balakrishnan, “Energyefficient communication protocol for wireless microsensor networks,” in Proc. 33rd Hawaii International Conference on System Sciences, January 2000. J. Werb and C. Lanzl, “Designing a positioning system for finding things and people indoors,” IEEE Spectrum, pp. 71-78, September 1998. P. Bahl and V.N. Padmanabhan, “User location and tracking in an inbuilding radio network,” Tech. Report MSR-TR-99-12, Microsoft Research, February 1999. N. Bulusu, J. Heidemann and D. Estrin, “GPS-less low cost outdoor localization for very small devices,” Tech. Report 00-729, Computer Science Department, University of Southern California, April 2000.

INTERNATIONAL WORKSHOP ON WIRELESS AD-HOC NETWORKS (IWWAN) 2004 [10] C. Savarese, J.M. Rabaey and J. Beutel, “Locationing in distributed adhoc wireless sensor networks,” in Proc. ICASSP2001, Salt Lake City, UT, May 2001. [11] L. Doherty, K. Pister and L. El Ghaoui, “Convex position estimation in wireless sensor networks,” in Proc. Infocom 2001, Anchorage, AK, April 2001. [12] D. Niculescu and B. Nath, “Ad hoc positioning system (APS),” Tech. Report DCS-TR-435, Rutgers University, April 2001. [13] J. G. Lim, K. L. Chee, H. B. Leow, Y. K. Chong, P. K. Sivaprasad and SV Rao, “Implementing a self-organizing wireless sensor network: experiences and challenges,” IEEE Conference on Mobile & Wireless Communications Networks 2003, Singapore, October 2003.

6

A Grid-Based Location Estimation Scheme using Hop ...

Report DCS-TR-435, Rutgers University, April 2001. [13] J. G. Lim, K. L. Chee, H. B. Leow, Y. K. Chong, P. K. Sivaprasad and. SV Rao, “Implementing a ...

320KB Sizes 5 Downloads 303 Views

Recommend Documents

Cover Estimation and Payload Location using Markov ...
Payload location accuracy is robust to various w2. 4.2 Simple LSB Replacement Steganography. For each cover image in test set B, we embed a fixed payload of 0.5 bpp using LSB replacement with the same key. We then estimate the cover images, or the mo

Cover Estimation and Payload Location using Markov ...
Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly ... Maximum a posteriori (MAP) inferencing:.

An Adaptive Network Coded Retransmission Scheme for Single-Hop ...
869. An Adaptive Network Coded Retransmission Scheme for Single-Hop Wireless Multicast Broadcast Services. Sameh Sorour, Student Member, IEEE, and Shahrokh Valaee, Senior Member, IEEE. Abstract—Network coding has recently attracted attention as a s

Hierarchical Location Management Scheme Based on ...
COMMUN., VOL.E87–B, NO.3 MARCH 2004. PAPER Special Section on Internet Technology IV ... tail stores, coffee shops, and so forth. Also, sensor networks.

Efficient Location-Aided Flooding Scheme Initiated by ...
for supporting various applications and protocols in wireless ad ... Many routing protocols for wireless ad ..... networking (monet) working group of the internet.

A User Location and Tracking System using Wireless Local Area ...
A User Location and Tracking System using Wireless Local Area Network. Kent Nishimori ... Area Network signal strength and Geographical. Information ..... The initial K-nearest neighbor algorithm [1] takes all of the K selected reference points and a

Inferring Residential Location Preferences Using a ...
Aug 24, 2010 - P6. P8. Increasing budget for housing ? (b) Reduced Choice/Chooser set after more privileged choosers have made their choice. Figure 1: A ...

Improving Location Prediction using a Social Historical ...
of social media where users are able to check-in to loca- tions they ... these predicted locations. Permission to ... [10] pro- posed the Order-k Markov model, which considers the fre- ... more recently visited places), and demonstrating how it can.

Muscle Location Optimization Using a GA
solutions found by the GA highlight those muscle groups that are most necessary for .... chosen as 1%. We used a selection criterion that used the top 5 parents from the ... either side of the skeletal model that give rise to mono- or bi- articulate 

Improving Location Prediction using a Social Historical Model with ...
Location-based Social Networks (LBSN) are a popular form of social media where users are able to check-in to loca- tions they have ..... [5] H. Gao, J. Tang, and H. Liu. gSCorr: modeling geo-social correlations for new check-ins on location-based soc

Inferring Residential Location Preferences Using a ...
Aug 24, 2010 - However, despite technological advances that have made ...... path. In principle, one could retain spatially coarse zones and ... of demographers and housing-career researchers can be made, and will result in a more reliable.

Distributed Fault Correlation Scheme using a Semantic ...
Semantic Publish/Subscribe system. Wei Tai, Declan ... hierarchical manager/agent topologies and rely upon significant ... The overall architecture of our fault management system ..... file encoded using SNMP4J and then included in a KBN.

A Generalized Data Detection Scheme Using Hyperplane ... - CiteSeerX
Oct 18, 2009 - We evaluated the performance of the proposed method by retrieving a real data ..... improvement results of a four-state PR1 with branch-metric.

A Generalized Data Detection Scheme Using ... - Semantic Scholar
Oct 18, 2009 - We evaluated the performance of the proposed method by retrieving a real data series from a perpendicular magnetic recording channel, and obtained a bit-error rate of approximately 10 3. For projective geometry–low-density parity-che

a novel pattern identification scheme using distributed ...
Jul 6, 2009 - macroblock temporal redundancy (ITR) (see Fig 1) which is static in successive frames. Indeed, few bits are used to signal zero residual error and zero motion for ITR to decoder, which is obviously significant when a sequence is encoded

A source authentication scheme using network coding ...
trade-offs with the widely accepted throughput benefits, especially in multicast scenarios. .... network coding can be used to provide security measures.

Phase estimation using a state-space approach based ...
In this paper, we propose an elegant spatial fringe analysis method for phase estimation. The method is presented in the context of digital holographic interferometry (DHI), which is a prominent optical technique for analyzing the deformation of a di

Joint Compression/Watermarking Scheme Using ...
Abstract—In this paper, a watermarking scheme, called ma- jority-parity-guided error-diffused block truncation coding. (MPG-EDBTC), is proposed to achieve high image quality and embedding capacity. EDBTC exploits the error diffusion to effec- tivel

inteligibility improvement using snr estimation
Speech enhancement is one of the most important topics in speech signal processing. Several techniques have been proposed for this purpose like the spectral subtraction approach, the signal subspace approach, adaptive noise canceling and Wiener filte

Two-hop secure communication using an untrusted relay
wireless communication systems from the physical layer up. Prominent such ... eavesdropper [8, 10, 17], and taking advantage of variations in channel state to ...

Two-hop secure communication using an untrusted relay - WCAN@PSU
wireless communication systems from the physical layer up. Prominent such examples include using multiple antennas to steer the transmitted ...... noise,” in Proceedings of the 62nd IEEE Vehicular Technology. Conference (VTC '05), vol.

Towards A Hip-Hop Aesthetic
Sep 8, 2006 - A QUICK RUNDOWN OF A FEW works that have been presented at the Hip- .... development money and a staff of designers and dramaturgs?

Protein location prediction using atomic composition ...
Dec 28, 2009 - subcellular localization and the second is the computational tech- nique employed for making prediction [1]. The biological features used for prediction include detection of protein sorting signal, ami- no acid composition, physiochemi