KR101934156B1 - A Fast and Energy-Efficient Asynchronous Neighbor Discovery Protocol for Cooperative Video Surveillance of Vehicles - Google Patents

Abstract

A fast and energy efficient asynchronous neighbor node discovery protocol for cooperative video surveillance in a vehicle is provided for discovery of neighbor nodes in opportunistic mobile networks, n _i the mode for selecting a parameter p _i corresponding to the target duty cycle (target duty cycle); A periodic wake-up mode in which the node n _i wakes up at a multiple of the selected prime number p _i ; And an active search mode for changing the node n _i to be activated to a consecutive slot c _i after a neighbor search request is received.

Description

A fast and energy efficient asynchronous neighbor node discovery protocol for cooperative video surveillance in a vehicle,

The present invention relates to an opportunistic network for cooperative video surveillance in a vehicle, and more particularly, to a neighbor node search method, which is an essential process prior to information exchange such as event notification, and more particularly, Efficient asynchronous protocol that searches for neighbors within a search range and reports quick and accurate search results.

Dashboard cameras (Dashcam) are becoming popular in order to collect evidence of traffic accidents on the road as well as to protect themselves from insurance fraud, as well as where many drivers are stationary.

However, since the dashboard camera has a limited viewing angle, it often occurs that the license plate of the vehicle or the driver's face related to various accidents of the parked car can not be recorded.

Accordingly, it is possible to consider cooperating with the dashboard cameras of the neighboring vehicles. By cooperating with the dashboard cameras of the neighboring vehicles, images can be obtained from various angles.

Collaboration of dashboard cameras In data transmission for a video surveillance network, neighbor node discovery is needed.

Recently, ad hoc networks such as opportunistic mobile networks have attracted much attention, and these networks can be applied to various services such as military, vehicle, medical, multimedia transmission have.

Conventional protocols, however, are designed to operate in such a manner that all nodes always search for all neighboring nodes regardless of whether a neighboring node needs to be searched. Therefore, when applied to an application such as a cooperative video surveillance of a vehicle, Consumption and a significant level of search delays.

A typical conventional protocol, a wake-up pattern of Disco, U-Connect, AARP and SearchLight is shown in FIG.

(A) is a wake-up slot, (b) is a U-Connect, (c) is an Adaptive Asynchronous Rendezvous Protocol (AARP) Pattern.

First, in the case of the Disco protocol, it is designed based on the Chinese-Remainder Theorem (CRT), where node n _i selects a pair of prime p _i1 and p _i2 , and node n _i wakes-up in each of the two prime numbers .

Referring to FIG. 1 (a), prime numbers 3 and 5 are selected as prime numbers p _i1 and p _i2 , and node n _i is maintained in a wake-up state in each multiple. 3, 5, 6, 9, 10 and 12 slots of the Disco protocol are 0, 3, 6, 9 and 12, wake-up state.

In the case of the AARP protocol, it is designed based on the CRT, and the node n _i selects the prime p _i and the node n _i , according to the conditions of n ≡ 0 (mod p _i ) or n ≡ 1 (mod p _i + 1) wake-up.

1 (b), the prime number 5 is selected as the prime number p _i , and 0, 5 for the first condition n ≡ 0 (mod 5), and the second condition n ≡ 1 (mod (5 + 1)) = 1 (mod 6), the 0, 1, 5, and 7 slots of the AARP protocol will maintain the wake-up state.

For U-Connect protocol, designed on the basis of CRT, node n _i is the selected prime number p _i, and node n _i is a prime number p _i p-sheet, as well as a multiple of _i ² Each slot is waked up in all slots from the first slot to (p _i +1) / 2.

Referring to FIG. 1 (c), a prime number 5 is selected as a prime number p _i and maintained in a wake-up state from the first slot to the third slot every 25 slots as well as a multiple of 5.

That is, 0, 5, and 10 for a multiple of 5, and 0, 1, and 2 for slots from the first slot 0 to (5 + 1) / 2 = 3 to the third slot 2, , 1, 2, 5, and 10 slots are maintained in the wake-up state.

SearchLight is not designed on the basis of CRT, so it does not need a prime number. The node n _i selects a number p _i to be used as a period and wakes-up in slot 0, which is the first slot in each period, and additionally wakes-up in the other slot. In SearchLight using continuous probing, the above additional wake-up slot is referred to as a probe slot. To select an additional wake-up slot, the node n _i uses a counter starting at 1, which increments by 1 for each period, then increases to p _i / 2, do.

Referring to FIG. 1 (d), the position of the probe slot is determined by a counter starting from 1, increases every cycle, ends at _pi / 2, and then starts at 1 again.

That is, n _i is 0 and 1 for the first period and 0 and 2 for the second period are wake-up (probe slots).

Although various protocols as described above are disclosed, it is designed to operate in such a manner that all the nodes always search for all neighboring nodes regardless of whether the neighboring nodes need to be searched. Therefore, when applied to applications such as cooperative video surveillance , Unnecessary energy consumption and a significant level of search delays.

Thus, the present inventor has sought to develop a protocol that is faster and more efficient than the protocols disclosed in the prior art.

M. Bakht, M. Trower, and R. H. Kravets, "Searchlight: Will not You Be My Neighbor ?," in ACM Mobicom'12, pp. 185-196, 2012. P. Dutta and D. Culler, " Practical Asynchronous Neighbor Discovery and Rendezvous for Mobile Sensing Applications, " in ACM Conf. on Embedded Networks Sensor Systems, pp. 71-84, 2008. A. Kandhalu, K. Lakshmanan, and R. Rajkumar, "U-Connect: A Low-Latency Energy-Efficient Asynchronous Neighbor Discovery Protocol," in Proc. IPSN'10, pp. 350-361, 2010. H. Ko, S. Oh, and C. Kim, "Adaptive, Asynchronous Rendezvous Protocol for Opportunistic Networks," Electronic Letters, vol. 48, no. 8, pp. 462-464, April 2012.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems.

Specifically, in opportunistic mobile networks, nodes have a different role in neighbor node discovery, searching for neighbors within a search range, and quickly and efficiently delivering protocols that are quicker and more accurate than previously disclosed protocols that report fast search results I would like to propose.

According to an aspect of the present invention, there is provided a method for discovering a neighbor node in opportunistic mobile networks, the method comprising the steps of: for _i selects the parameters p _i corresponding to the target duty cycle (target duty cycle) mode; Wherein the node n _i is wake-up in a multiple of the selected parameter p _i ; And an active search mode in which the node n _i changes to be activated as a consecutive slot c _i after a neighbor search request is received.

It is also preferred that half of the target duty cycle is used in the periodic wake-up mode and the other half is reserved for the active search mode.

Also, the active search mode is preferably performed when a wake-up ratio (WR) is established according to the following equation using the following equation.

WR < / RTI > 2 / _pi where WR is the ratio of wake-up slots for the entire slot and pi is the randomly selected parameter.

Also, the active search mode is preferably performed after the delay until the WR is satisfied in the above equation when WR is not satisfied in the above equation.

The apparatus may further include a mode for returning to the periodic wake-up mode after the active search mode.

Also, when the active search mode is performed as frequently as possible in multiples of _pi , it is preferable that the number of slots between the start points of two consecutive active search modes can be calculated by the following equation.

Equation, c _i is the search of the node n _i in a continuous slot, p _i is a randomly selected parameter.

Also, the mode for selecting the parameter p _i comprises: selecting a target duty cycle d (d _low <d ≤ d _high ); Selecting an inverse number of a parameter closest to a half of the selected target duty cycle; And using the selected parameter to specify a search range in a neighbor search request.

In addition, the parameter is preferably a prime number.

Node n search range of _i c _i may be a specified value to the neighboring search request, the node n _i if there is no request, the target half of the duty cycle (1 / p _i) as close as possible, to maintain the wake-up ratio in the have.

When the wake-up ratio is less than 2 / p _i when the neighbor search request is received, the node n _i is more accurate than the asynchronous neighbor search protocol for all neighbor n _{k in} the case of p _k ≤ c _i Can be obtained.

In an application such as a cooperative video surveillance network of a dashboard camera, there is no need to search for a neighboring node while recording a video. Therefore, in an application such as a vehicle accident, In the case of a search request, the FEND according to the present invention is much superior to the conventional protocol in terms of search delay time and energy consumption.

The slot size used in the simulation is 10ms, which is acceptable for low-cost communication equipment.

Figure 1 shows a prior art wake-up pattern.
2 shows the concept of FEND according to the present invention.
FIG. 3 shows a wake-up schedule (or pattern) when p _i = 7 and c _i = 11 in FEND according to the present invention.
4 shows a change in the operation mode of the FEND according to the present invention.
Fig. 5 shows the wake-up ratio, the event-based search delay average, and the WR-delay product when nodes with a target duty cycle of 5% in the symmetric case search for adjacent nodes having the same target duty cycle.
FIG. 6 shows the event-based search delay average and WR-delay product when a node having a target duty cycle of 2% in the asymmetric case searches for adjacent nodes having a target duty cycle of 5% and 11%.
FIG. 7 shows the event-based search delay time and the WR-delay product when a node with a target duty cycle of 11% in the asymmetric case searches for adjacent nodes having 2% and 5% target duty cycles.

Hereinafter, the fast and energy efficient Asynchronous Neighbor Discovery Protocol of the present invention is referred to as " FEND " for convenience of explanation.

The time used in " FEND " is divided into fixed width slots, successive time slots are represented by continuous integers, and all nodes start counting the time slot progression by the number of slots under the assumption that start times may be different.

Also, the " periodic wake-up mode " means an operation mode in which a user or an application periodically wakes up before a neighbor search request is received.

The " active search mode " means an operation mode in which a neighbor search request is activated after a user or an application receives the neighbor search request.

Also, the " search delay time " means a delay time between when a node receives a neighbor search request and when it finds a neighbor node.

In addition, " symmetrical case " means that all nodes have the same duty cycle, and " asymmetrical case " means the remaining cases except for the symmetric case.

Also, p _i is an arbitrarily selected parameter.

Since p _i is a parameter, it is not limited to the dimension, and a preferable example is an integer. In the present invention, a prime number is used for the performance comparison with the conventional technique.

Definition of FEND

The FEND of the present invention includes a mode in which a node n _i selects a parameter p _i according to a target duty cycle, a periodic wake-up mode in which a wake-up occurs in a multiple of a selected parameter p _i , _i to be activated to the continuous slot c _i , and a mode to return to the periodic wake-up mode after performing the active search mode.

Node n _j in an active search mode and wake-up slots in a row c _j, c _j consecutive slots are specified in the parameters of the search range of the neighbor discovery request by the node n _j of the user or an application with a number.

In active search mode, node n _j can search all neighboring nodes n _k that wake up in multiples of p _k (p _k ≤ c _j ).

A preferred embodiment of the FEND will be described in detail with reference to FIG. In FIG. 2, the parameter p _i is the wake-up interval of the node n _i , and the continuous slot c _i specifies the search range of the user or application of the node n _i (the gray slot is the wake-up slot).

Node n _i is wake-up from a multiple of the parameter _{p i (1 ≤ i ≤ 4} ), p 1 = 5, p 2 = 7, p 3 = 11, p 4 = when 13, node n ₅ is c ₅ = 11 in a continuous slot c ₅ .

Node n ₅ can search for neighboring n ₁ , n _2, and n ₃ wake-up in multiples of p ₁ , p _2, and p _{3 that} are less than or equal to search range c ₅ .

Another preferred embodiment of FEND will now be described in more detail with reference to FIG. FIG. 3 shows a wake-up schedule (or pattern) when p _i = 7 and c _i = 11 in FEND.

Also, in FIG. 3, L _A is the number of slots in the active search mode and L _W is the number of slots in the periodic wake-up mode.

l is the number of slots between the first wake-up slot of the periodic wake-up mode from the end of the active search mode.

The contiguous slot c _i is a search range parameter of node n _i defined by the user or application and p _i is a parameter that creates a wake-up schedule (or pattern) based on the target duty cycle of node n _i .

FEND includes two modes: periodic wake-up mode and active search mode.

In Figure 3 (a) refer to To explain in detail, the periodic wake-up mode, the node n _i is wake-up from a multiple of the parameter p _i, the parameter p _i is based on the target duty cycle of the node n _i .

Node n _i in the active search mode is specified in the search range of the continuous slot in the c _i and wake-up, the continuous slot of the node n _i c _i is the user or the application of a neighbor discovery request (see Fig. 3 (b)).

At FEND, node n _i starts periodic wake-up in periodic wake-up mode and changes its operating mode to active search mode whenever it receives a request from user or application.

Further, at the end of the active search mode, the operation mode is automatically returned to the periodic wake-up mode (see Fig. 3 (c)).

In FEND, half of the target duty cycle is used as the periodic wake-up mode and the other half is defined as reserved for the active seek mode.

Thus, the node n _i selects the inverse of the parameter p _i such that it is as close as possible to half of the target duty cycle.

With respect to the change condition of the operation mode, the node n _i can start the active search mode whenever WR (Wake-up Ratio) is 2 / p _i or less.

If WR is not less than 2 / p _i , node n _i should wait until WR is equal to or less than 2 / p _i .

3 (c), assuming that the active search mode is activated as frequently as possible for a multiple of p _i , the slot between the active search mode and the next active search mode is given by Equation (4) (4) will be described in detail in the proof part to be described later.

This will be described in detail with reference to FIG.

FIG. 4 shows an example of an operation mode change in FEND, in which the active search mode is activated as often as possible, so that WR at t ₁ , t ₄ and t ₈ represents 2 / p _i , (See Fig. 4 (a)).

Therefore, as shown in equation (4), L _A + L _W

.

In the case of WR = 2 / p _i at t ₁ , the neighbor search request arrives at t ₂ , t ₃ and t ₇ (see FIG. 4 (b)).

Since at t ₂ WR = ₂ / p _i and an active search mode can be activated at t _2, adjacent search result is reported to the user or application to t ₂ + _A L.

t ₃ and t ₇ because the WR> is 2 / p _i, the active search mode, and t ₃ can not be activated at t _7.

Therefore, the start of the active search mode must be delayed until t ₄ and t _8, respectively. In this case, the results are reported as t ₄ + L _A and t ₈ + L _A.

In the case of t ₁ to WR = 2 / p _i, the neighbor discovery requests to t ₅ and t ₆ arrives. the active search mode is activated at t ₅ and t ₆ , since t R ₅ and t ₆ are WR <2 / p _i , resulting in t ₅ + L _A and t ₆ + L _A reported (see FIG. ).

Specifying the search scope of FEND

Hereinafter, the search range designation of the FEND will be described in detail.

When all nodes select their duty cycle according to the remaining battery level d _low and d _high , set the search range's default value as close as possible to the half of d _low .

A method for an application of a node to discover nodes with a high amount of residual energy is to select a target duty cycle d (d _low <d ≤ d _high ), and determine whether the reciprocal of the parameter is closest to the half of the selected duty cycle And using the selected parameter to specify the search range in the neighbor search request.

The neighbor search request may include multiple search ranges and may include the minimum number of neighboring nodes to be searched.

For example, the case that contains the least number of neighbors will be the neighbor discovery request, the search node and a ≤ c ₁ c ₂ c ₁ of the two search ranges, and c ₂ are provided.

In this case, the node first tries to wake up continuously in consecutive slot c ₁ after changing the mode of operation to active search mode first. If it finds more nodes than the minimum number of neighbor nodes to be searched, the node immediately returns to the periodic wake-up mode.

If the neighboring nodes have not been sufficiently searched in the consecutive slot c ₁ , the node tries to continue to wake up in the next consecutive slot c ₂ -c ₁ for the neighbor search.

In high-density networks, this approach not only reduces energy consumption due to premature termination of the active search mode, but also shortens the event-based search latency because rapid activation of the active search mode allows for reduced energy consumption.

Performance evaluation

We evaluate the FEND, AARP, Disco, and U-Connect according to the present invention with the following criteria through simulations using the ns-3 (Network Simulator 3) program: wake-up ratio (WR) Time, WR-delay product.

We also use three target duty cycles of 2%, 5%, and 11% in the simulation.

[Table 1]

FEND, AARP, Disco, and U-Connect are based on the duty cycle. The FEND, AARP, Disco, and U-Connect are based on a small number of values used in FEND, AARP, Select a prime number.

Again, the WR for AARP, U-Connect and Disco is determined by the chosen prime.

However, the WR obtained by the parameter p _i selected by the node n _i in FEND varies between 1 / p _i and 2 / p _i according to the schedule (or pattern) of the neighbor search request.

Again, it is preferable that the dimension of the parameter pi of FEND is not limited, but the parameter pi of FEND selects a prime number to obtain the performance evaluation result under the same condition.

The time interval at which neighbor node discovery requests arrive at all nodes follows the exponential distribution average value.

When the dashboard camera starts to shoot a video after searching for its neighbors, it is not necessary to search for the neighbor again before terminating the video shooting.

Therefore, if a dashboard camera records 30 seconds of video due to an event, μ in a cooperative video surveillance network of a dashboard camera will obviously be greater than 30 seconds.

However, in order to analyze the performance of FEND for various μ, simulation sets the range of μ from 5 seconds (500 slots) to 100 seconds (10,000 slots). The slot size used in the simulation is 10ms, which is acceptable for low-cost communication equipment.

To obtain reliable results, 100 experiments were repeated for one μ, and one experiment was performed for 30,000 seconds (3,000,000 slots).

In each experiment, the starting slot of each node is randomly selected between 0 and 100. [

In order to evaluate the performance of FEND, we make the node n _i search for the node n _j in two extreme situations: In the first case, the node n _j represents the search range c _{j (} c _j = p _{j )} Use it as a parameter to enable active navigation mode as often as possible.

In the second case, node n _j does not use active search mode at all.

For the latter and the former, they are called FEND-W and FEND-P, respectively.

First, the case of symmetry is evaluated. The simulation of the network consists of two nodes, n ₁ and n ₂ , and the target duty cycle of both nodes is equal to 5%. Each protocol will select a small number, as shown in Table 1, and in the case of symmetry is the navigation of the node n ₁ in FEND because the range c ₁ is the same as p _1, nodes n attempt to duty cycle ≥ 5% neighborhood of the target of ₁ .

In FIG. 5, a wake-up ratio of FEND and conventional protocols, an event-based search delay average, and a WR-delay product are shown in the case of symmetry, and therefore, detailed evaluation will be made with reference to FIG.

5 (a) shows a graph of the node n ₁ WR attempting to traverse the node n ₂ in the case of symmetry is shown.

The WR of AARP, Disco, U-Connect is determined by the selected decimal number (see Table 1), but the WR of FEND depends on the neighbor search request time interval.

Using equation (4), since c ₁ = p ₁ = 41

= 1640 slots (slot size 10 ms to 16.4 seconds).

Therefore, in FEND, the WR of node n ₁ is 2 / p ₁ at μ ≤ 15 seconds, and it decreases at μ> 15 seconds according to the average arrival time.

FIG. 5 (b) is a graph showing the average of the event-based search delay time of the node n ₁ searching for the node n ₂ in the symmetric case.

It can be seen that the average of the event-based search latency of AARP, Disco, and U-Connect is about 3.91 seconds to about 5.37 seconds regardless of the neighbor search request time interval.

The average value of the event-based navigation delay of AARP, Disco and U-Connect is considered to be too large for cooperative video surveillance of vehicles.

In FEND, the start of the active search mode may be delayed, as shown in FIG. 4, in the case of μ <15 seconds.

Therefore, at μ <15 seconds, the average of the event-based search latency of FEND-P and FEND-W is not small compared to AARP, Disco and U-Connect.

In addition, since the number of wake-up slots of n ₂ in FEND-W is only half that of FEND-P, the average of the event-based search delay time of FEND-W is longer than that of FEND-P.

However, in the case of μ> 15 seconds, since the active search mode can be started immediately as shown in FIG. 4 (c), the event-based search delay time average in FEND-P and FEND-W is about 0.20 seconds .

Unlike conventional protocols, the event-based search delay time average in FEND-P and FEND-W is considered to be sufficiently small for cooperative video surveillance in vehicles.

FIG. 5 (c) is a graph showing the WR-delayed product of node n ₁ searching for node n ₂ in the symmetric case. 5 (c) has a shape similar to that of FIG. 5 (b), but the result of considering both energy consumption and delay time is shown.

WR of node n ₂ in FEND-W is half of FEND-P, and in the case of μ> 15 seconds as shown in FIG. 5 (a), WR of node n ₁ gradually decreases.

Let R _{i, j be the} ratio of the WR sum of n _i and n _j in FEND-W for the WR sum of n ₁ and n ₂ in FEND-P, then R _1,2 value is 0.75 for μ ≤ 15 seconds , μ> 15 seconds, the value of R _1,2 continuously decreases as the value of μ increases.

Thus, at μ ≤ 15 seconds, the difference between the WR-delay products of FEND-P and FEND-W is less than the difference of the event-based search delay time averages of FEND-P and FEND-W.

In the case of μ> 15 seconds, the WR-delay product of FEND-W becomes smaller than the WR-delay product of FEND-P.

As a result, at W = 100 seconds, the WR-delay product of FEND-W is only about 1 / 36.5 times the WR-delay product of AARP, and is only about 1 / 46.6 times that of Disco's WR-delay product.

Next, the asymmetric case protocol is evaluated. The simulation of the network consists of three nodes n ₁ , n ₂ , and n ₃ , with target duty cycles of 2%, 5%, and 11%, respectively, and a prime number is selected as shown in Table 1.

The evaluation of the asymmetric protocol will be described in detail with reference to FIG.

FIG. 6 shows a graph of FEND versus event-based search delay average and WR-delay product of conventional protocols in the asymmetric case.

Figure 6 (a) and in (b) when the asymmetric, the target duty cycles ≥ 5% of the detected target node, c ₁ = 41 a, the node n event based on the node n ₁ for searching a ₂ and n ₃ search latency The average is shown.

Using the equation (4), it is found that p ₁ = 89 and c ₁ = 41

= 3,560 slots (35.6 seconds with a slot size of 10 ms).

Therefore, at μ = 30 sec, the event-based search delay time average of FEND-P and FEND-W is not much shorter than that of AARP, Disco, and U-Connect.

However, in Fig. 6 (a), the event-based search delay time average of FEND-P and FEND-W is only about 0.25 seconds in the case of μ> 35 seconds. On the other hand, , Which is not suitable for cooperative video surveillance of the vehicle, from about 7.26 seconds to about 7.57 seconds.

Similarly, in FIG. 6 (b), the event-based search delay time average for FEND-P and FEND-W is only about 0.11 seconds, , Which is not suitable for cooperative video surveillance of the vehicle, is about 2.87 seconds to about 3.48 seconds.

Figure 6 (c) and (d) the asymmetrical case, the target duty cycle in ≥ 5% of the found nodes, namely c ₁ = 41, node n WR- delay product of the nodes n ₁ and n ₃ for searching a _second Respectively.

If, as previously mentioned, called the WR agreement ratio of n ₁ and n ₂ of the FEND-W for the WR sum of n _i and n _j at FEND-P R _{i, j,} R _1, if μ ≤ 35 seconds _{, The} value of ₂ is 0.66, the value of R _1,3 is 0.58, and when μ> 35 seconds, the values of R _1,2 and R _1,3 continuously decrease as the value of μ increases.

6 (c) and (d), the WR-delay product of FEND-W is always smaller than the WR-delay product of FEND-P. Among the conventional protocols, Disco has the lowest WR-delay product. Compared with Disco, at W = 100 seconds, the WR-delay product of FEND-W is larger than the WR- And only about 1 / 62.2 in FIG. 6 (d), and about 1 / 40.0 in FIG. 6 (d).

In case of asymmetry in FIGS. 7 (a) and 7 (b), the event-based search delay time average of node n ₃ searching nodes n ₁ and n ₂ with target duty cycle ≥ 2% and c ₁ = .

Using the equation (4), at p ₃ = 17 and c ₃ = 89, L _A + L _W =

= 1,411 slots (14.11 seconds with a slot size of 10 ms).

Therefore, at μ = 10 sec, the event-based search delay time average of FEND-P and FEND-W is not much shorter than that of AARP, Disco, and U-Connect.

However, in Fig. 7 (a), the event-based search delay time average in FEND-P and FEND-W is only about 0.44 seconds in the case of 占 15 15 seconds, , Which is not suitable for the cooperative video surveillance of the vehicle, from about 2.86 seconds to about 3.49 seconds.

Similarly, in Fig. 7 (b), the event-based search delay time average for FEND-P and FEND-W is only about 0.20 seconds, , Which is not suitable for cooperative video surveillance of the vehicle, is about 1.33 seconds to about 1.50 seconds.

Figure 7 (c) and (d) the asymmetrical case, the duty cycle of the node in the found target ≥ 2%, or _1, c = 89, node n WR- delay product of the nodes n ₁ and n _{2 3} navigating Respectively.

As mentioned above, when the WR agreement ratio of n _i and n _j at FEND-W for at FEND-P in WR sum of n ₁ and n ₂ R _{i, j} la, R ₁ in the μ if <15 seconds _{, The} value of ₃ is 0.92, the value of R _2,3 is 0.85, and when μ ≤ 15 seconds, the values of R _1,3 and R _2,3 continuously decrease as the value of μ increases.

7 (c) and (d), the WR-delay product of FEND-W is always smaller than the WR-delay product of FEND-P. Among the conventional protocols, Disco has the lowest WR-delay product. Compared with Disco, at W = 100 seconds, the WR-delay product of FEND-W is larger than the WR- Not only about 1 / 11.5 but also about 1 / 11.9 in Fig. 7 (d).

As described above, when the video of the dashboard camera records 30 seconds when triggered by the event, the arrival time interval of the neighbor search request is, of course, greater than 30 seconds.

5 to 7, it can be seen that FEND exceeds the existing protocols in terms of search delay time and energy consumption at &lt; RTI ID = 0.0 &gt;

[proof]

L _A is the number of slots in active search mode, and L _W is the number of periodic wake-up slots.

Also, l is the number of slots between the end of the active search mode and the first wake-up slot of the periodic wake-up mode thereafter, and c _i and p _i are parameters.

Also, any number k has a range of k? 0.

Equation (1)

Here, the ratio of wake-up slots to L _A + L _W should be equal to 2 / p _i .

Equation (2)

In the equation (2)

Equation (3)

From equations (1) and (3)

Equation (4)

According to equation (4), when the active search mode is activated as often as possible in multiples of _pi , the number of slots between the start points of two consecutive active search modes is

As shown in FIG.

While the present invention has been described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims. It will be appreciated that embodiments are possible. Accordingly, the scope of protection of the present invention should be determined by the claims.

Claims (8)

A method for discovery of neighbor nodes in opportunistic mobile networks, the method comprising:
(a) selecting a parameter p _i according to a target duty cycle of a node n _i of the opportunistic mobile network;
(b) performing a wake-up mode in which the node n _i is periodically waked up in a multiple of the selected parameter p _i ; And
(c) performing an active search mode to change the node n _i to be activated to a consecutive slot c _i after a neighbor search request is received,
Half of the target duty cycle is used in the periodic wake-up mode, the other half is reserved for the active search mode,
The active search mode is performed when a wake-up ratio (WR) is established according to the following equation using the following equation, and when WR is not satisfied with the above equation, Lt; / RTI &gt; is performed after the delay,
WR? 2 / p _i
In the equation, WR is the ratio of wake-up slots to total slots,
A method for searching neighboring nodes in an opportunistic mobile network using an asynchronous neighbor discovery protocol.
delete delete delete The method according to claim 1,
After the step (c)
Returning to step (b)
A method for searching neighboring nodes in an opportunistic mobile network using an asynchronous neighbor discovery protocol.
The method according to claim 1,
If the active search mode is performed as often as possible in multiples of the parameter _pi ,
The number of slots between the start point of two consecutive active search modes can be calculated by the following equation,

In the equation, c _i is the search range of the node n _{i in} the continuous slot,
A method for searching neighboring nodes in an opportunistic mobile network using an asynchronous neighbor discovery protocol.
The method according to claim 1,
The step (a)
(a1) selecting the target duty cycle d (d _low <d? d _high );
(a2) selecting an inverse of the parameter closest to half of the selected target duty cycle; And
(a3) using the parameter to specify a search range in a neighbor search request,
A method for searching neighboring nodes in an opportunistic mobile network using an asynchronous neighbor discovery protocol.
8. The method of claim 7,
The parameter is a prime number,
A method for searching neighboring nodes in an opportunistic mobile network using an asynchronous neighbor discovery protocol.

Images