Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Bloom filter for fixed-size Beacon in VANET Project description Introduction VANETs or vehicular ad hoc networks are the spontaneous creation of a wireless network for data exchange – to the domain of vehicles. They are a main component of intelligent transportation systems. The advantage of VANET that nodes can directly communicate with each other. Consequently, most work in these networks concentrates on efficient data delivering. Mechanism of delivering need a beacon to discover neighbors and share local data. Beacons - packets that are periodically broadcast to maintain integrity of network. Such packets need to transmit a lot of necessary information like source ID, neighbors’ ID, packets’ ID, etc. This leads to some serious problems. Description of the problem One of the main problem is limited resources in wireless networks. Many approaches exist to avoid this problem. 1) Reducing the frequency of beacon broadcasting. 2) Reducing the number of beacons. 3) Desynchronization methods to prevent the collision of beacon broadcasting. These approaches are useful to solve contention problem, but all of them affects the accuracy and volume of beacon information. Size of beacon depend on information inside it. For instance, beacons in topology-based protocol require an 1-hop neighbor identifier list and packet identifier list as acknowledgments for reliable data dissemination. The beacon structure can be illustrated in Fig. 1. Source ID Neighbor ID #1 Neighbor ID #2 ... Neighbor ID #n Packet ID #1 Packet ID #2 ... List of 1-hop Neighbor IDs Packet ID #n List of packet IDs Fig. 1. Beacon structure. As result, the beacon size will increase when a node has more neighbors. Moreover, in many cases beacon need to transmit 2-hop neighbor identifier list too. This problem often is not considered, because it require additional space and can lead to deterioration of connection. To solve these problems we use Bloom filter to store packet identifier list and 1-hop, 2-hop neighbor identifier list. Motivation Today a lot of applications and services are used for a safe and convenient driving: a navigation system, traffic management system, collision avoidance system, etc. All of these applications require an efficient way to collect and exchange information, so the problem of limited resources is actual. Our motivation to solve this problem by using Bloom filters and creating fixed-size beacon. It will help the networks be more scalable and beacon will not affect the data exchange. Existing solutions and their limitations A Bloom filter is simple structure that was designed for reducing memory usage in limited resource systems. In initial state Bloom filter contain a bit array that all bits set to “0”. When new element is added, it will be fed different hash functions. These functions compute position in bit array that will be set to “1”. Bloom filter structure can be illustrated in Fig.2. Fig. 2. Bloom filter structure. Kulit Na Nakorn, Yusheng Ji and Kultida Rojviboonchai have already proposed efficient beacon structure with Bloom Filter. This structure is shown in Fig.3. Source ID List of elements Bloom Filter Source ID Neighbor ID #1 Neighbor ID #2 ... Neighbor ID Packet ID #1 Packet ID #2 ... Packet ID #n #n List of 1-hop Neighbor IDs List of packet IDs Fig. 3. Fixed-size beacon structure. This solution has one important disadvantage. Bloom filter never gives false negative errors, but the result can be a false positive. Suppose that m – number of bits in a Bloom Filter array, n – number of elements in the set and k – number of hash functions. The probability of false positive results can be calculated by (1). 𝑝 = (1 − (1 − 1 𝑘𝑛 𝑘 ) ) 𝑚 (1) Since this solution uses one Bloom filter to store both list of neighbors’ ID and list of packets’ ID the probability of false positive errors increases. To improve this solution and solve the problem with 2-hop neighbors we designed beacon structure that can be illustrated in Fig.4. List of Neighbor IDs Source ID List of packet IDs Bloom Filter Bloom Filter Source ID Neighbor ID #1 Neighbor ID #2 ... Neighbor ID Packet ID #1 Packet ID #2 ... Packet ID #n #n List of 1-hop and 2-hop Neighbor IDs List of packet IDs Fig. 4. New fixed-size beacon structure. We use two separate Bloom filters in beacon structure. First filter store packets’ identifiers which are used by protocol for missing packet detection. Second filter contains list of 1-hop and 2-hop neighbors. 2-hop neighbors’ identifiers are represented as result of exclusive or operation with the source ID and the ID of intermediate vehicle. False positive error results The optimal number of hash functions was calculated by (2). 𝑚 ln 2 (2) 𝑛 Since we solve problem with neighbor detection Bloom filter with nodes’ identifiers is tasted. The identifiers are representing like 32-bit numbers. Table 1 shows the false positive error rates and Bloom filter configurations. 𝑘= Table 1. Bloom filter configuration False positive error rate 5% 10% 15% 20% 25% No. of hash functions No. of bits 4 886 3 681 3 561 2 476 2 410 Number of elements, number of bits in array and number of hash functions initializes Bloom filter. The given number of elements is generated like list of neighbors. Then each neighbor from list is added to the Bloom filter. After this filter is randomly queried with elements that are and are not in the set. Testing was carried out on a machine with the characteristics shown in table 2. Table 2. Characteristics of machine CPU RAM System type Intel(R) Core(TM) i5-5200U 2.20 GHz 6.00 GB 64-bit operating system, processor x64 The false positive error results are shown in table 3. Table 3. False positive error results Number Number Number of Theoretical Averege of bits of hashes elements error Worst case case Best case 886 4 250 20% 24% 20,9% 18% 200 12% 17% 12,4% 10% 150 5% 10% 5,3% 2% 681 3 561 3 476 2 410 2 100 50 10 250 200 150 100 50 10 250 200 150 100 50 10 250 200 150 100 50 10 250 200 150 100 50 10 1% 0% 0% 29% 20% 11% 4% 0% 0% 40% 28% 16% 7% 1% 0% 42% 32% 21% 11% 3% 0% 49% 38% 26% 14% 4% 0% 3% 0% 0% 33% 24% 18% 9% 4% 0% 50% 36% 20% 11% 4% 0% 47% 39% 28% 18% 6% 0% 52% 43% 29% 18% 8% 0% 1,6% 0,0% 0,0% 29,5% 20,6% 11,7% 4,7% 1,0% 0,0% 40,4% 28,9% 16,2% 7,6% 1,3% 0,0% 41,4% 32,3% 20,2% 10,0% 3,6% 0,0% 48,6% 38,2% 26,2% 14,4% 3,8% 0,0% 0% 0% 0% 27% 18% 7% 3% 0% 0% 36% 27% 12% 4% 0% 0% 36% 28% 14% 5% 2% 0% 41% 35% 24% 11% 0% 0% Comparison analysis Theoretical, the worst, average and the best false positive error results are shown in Fig. 5-8. 50% 45% 40% Error rate (%) 35% 30% 5% 25% 10% 20% 15% 20% 15% 25% 10% 5% 0% 0 50 100 150 200 250 300 Number of elements Fig. 5. Theoretical false positive error. 50% 45% 40% Error rate (%) 35% 30% 5% 25% 10% 20% 15% 20% 15% 25% 10% 5% 0% 0 50 100 150 200 250 Number of elements Fig. 6. The worst-case false positive error. 300 50% 45% 40% Error rate (%) 35% 30% 5% 25% 10% 20% 15% 20% 15% 25% 10% 5% 0% 0 50 100 150 200 250 300 Number of elements Fig. 7. Average case false positive error. 50% 45% 40% Error rate (%) 35% 30% 5% 25% 10% 20% 15% 20% 15% 25% 10% 5% 0% 0 50 100 150 200 250 300 Number of elements Fig. 8. The best-case false positive error. The false positive error results for beacon structure with one Bloom filter are shown in Fig. 9. Obviously, results also depend on types of hash functions, machine settings, etc. Comparison analysis shows that in average case proposed Bloom filter gives better or worse results depending on different filter settings. In additional our Bloom filter contain information about 2-hop neighbors. This property can be useful for many applications. Fig. 9. The false positive error results for beacon structure with one Bloom filter. Conclusions New structure of fixed-size beacon was designed and tasted in this project. Proposed beacon allows to transmit necessary data and additional information about 2-hop neighbors in efficient way. Advantage of this structure that volume of data do not effect on beacon size. It can solve problem with limited resources, which was described in the beginning. References [1] [2] K. N. Nakorn, Y. Ji and K. Rojoviboonchai, “Bloom filter for fixed-size beacon in VANET”, Proc. IEEE Vehicular Technology Conference (VTC Spring), 2014 G. Samara, A. O Abu Salem and T. Alhmiedat, “Pover control protocols in VANET”, Proc. European Journal of Scientific Research, ISSN 1450-216X / 1450-202X Vol. 111 No 4 September, 2013, pp.571 – 576. http://www.europeanjournalofscientificresearch.com