CN101835235B - Routing method for heterogeneous network based on cognition - Google Patents

Abstract

The invention discloses a routing method for a heterogeneous network based on cognition, mainly solving the problem that the existing method considers only part of environment factors of the network, and neglects the dynamic complexity of a heterogeneous network environment so as not to cognize and predict the environment. The method comprises the following steps: firstly, judging the types of the existing available networks, perceiving a network environment state, evaluating the network environment state by using a Q-learning method and calculating end-to-end time delay and the estimated values thereof; secondly, using a utility function based on multiple parameters to obtain the utility values of various networks according to the perceived network environment state; then selecting the network with the biggest utility value by a probability of P=0.9, and selecting other available networks as target networks by the probability of 1-P; and finally continuously transmitting businesses, implementing network switch and updating the network environment state by the target networks. The routing method improves the utilization ratio of network resources, achieves load balance of the network and can be used in the heterogeneous network environment.

Description

Routing method for heterogeneous network based on cognition

Technical field

The invention belongs to wireless communication technology field; Relate to method for routing; A kind of specifically in heterogeneous network based on the routing method for heterogeneous network of cognition, be used for the wireless network portable terminal and select a suitable path transport service in the overlapping region covered of multiple network.

Background technology

Along with the continuous development of the communication technology, because present frequency spectrum resource all is a fixed allocation, the available spectrum resource is fewer and feweri in the entire spectrum space, and it has been one of present problem demanding prompt solution that frequency spectrum resource distributes inequality.Since the current network environment receive multiple factor such as frequency spectrum resource in short supply, electric wave transmission characteristic aspects such as constraint restriction and can't continue expansion, therefore need a kind of new technology that can make full use of the resource of diverse network and guarantee the raising of network service quality of research.

The trend of future network is progressively to move towards cognitiveization, syncretization, ubiquitousization, i.e. cognition network.Cognition network is to possess one through perception current network condition and make plan and decision-making, thereby makes the network system of the cognitive process of further action according to condition, can learn and predict network environment and make decision-making subsequently with this.Simultaneously, it also is future network Development Trend place that multiple network merges, and the heterogeneous network fusion has many-sided advantage, as improving network scalability, makes full use of Internet resources, better meets user's request etc.Therefore, need a kind of rational routing method for heterogeneous network of design to solve above problem, promptly in the overlapping covered network user's transport service of all hoping to select an optimum active user of multiple network.Have under the heterogeneous network environment of cognitive ability; How rationally to utilize both sides' demand of Internet resources and As soon as possible Promising Policy user and network; How the user selects a suitable network transport service through the cognitive ability of self; Promptly carrying out network and switch, to reach the raising of network performance, is the difficult point place of route selection algorithm.

Occur at present much about the heterogeneous network Study on selection method, existing now algorithm mainly is based on the algorithm of fuzzy logic, based on some factor such as RSS; Power; The algorithm of QoS demands such as time delay, based on the network selecting algorithm of effectiveness, and based on the decision-making technique of confirming parameter weighting etc.These algorithms have only been considered the component environment factor of network; Do not consider the interaction between user preferences and user type and the network environment; And ignored the environmental characteristics of the DYNAMIC COMPLEX of heterogeneous network; The well complicated isomerism and the dynamic change property of adapted to isomerous network, and can not carry out knowledge and anticipation to network environment, can't reach the requirement of future network development trend.

Summary of the invention

The object of the invention is to overcome the deficiency of above-mentioned prior art; A kind of routing method for heterogeneous network based on cognition is provided; Complexity and dynamic change property with adapted to isomerous network; Take all factors into consideration the interaction between user preferences and user type and the network environment, environment is carried out knowledge and anticipation, satisfy the demand of future network development.

For realizing above-mentioned purpose, method for routing of the present invention comprises the steps:

(1) mobile terminal period property obtain current available network type, the sensing network ambient condition adopts the Q-learning method that the network environment state is estimated to obtain the estimated value of each network environment state, and obtains end-to-end time delay and estimated value thereof;

(2) portable terminal is according to the network environment state that obtains, to network availability bandwidth f _Cap, Internet utilization fee uses f _PricWith end-to-end time delay T _TotalCarry out normalization and handle, obtain the weighted value that end-to-end time delay, network availability bandwidth and Internet utilization fee are used, and, select the maximum network of value of utility through following functional expression:

Max?U _i＝F _i(f _cap，f _pric，T _total)

Wherein, F _i(f _Cap, f _Pric, T _Total)=ω ₁(BW _Max-f _Cap)+ω ₂f _Pric+ ω ₃T _TotalBe based on the utility function of multi-parameter, ω ₁Be the weighted value of network residual available bandwidth, ω ₂Be the weighted value that Internet utilization fee is used, ω ₃Be the weighted value of end-to-end time delay, above weighted value adopts analytic hierarchy process (AHP) to try to achieve;

(3) according to the maximum network of the value of utility that obtains, portable terminal is selected this network with probability P=0.9, selects other available network with probability 1-P, as the objective network that is used for professional transmission;

(4) portable terminal carries out the business transmission according to the objective network that obtains, and judges whether this objective network is identical with current network, if identical, then portable terminal directly uses this Network Transmission professional; Otherwise; Portable terminal sends handoff request, switches to the objective network from current network, and business is continued transmission through objective network; Upgrade the estimated value of network environment state and each state simultaneously; When business was sent in the backbone network, portable terminal was also made the same behavior to the node in the backbone network, to guarantee the optimization of service transmission path.

The present invention compared with prior art has following advantage:

1) the present invention is owing to utilize the cognitive ability to environment at portable terminal in professional transmission course; The network environment state is predicted and estimate; The ambient condition situation that makes portable terminal obtain is approached truth more; Rather than believe the historical information of network environment state simply, and can predict the network ambient condition trend of following a period of time, help portable terminal and make the selection that meets the Resource Allocation in Networks situation; Reduced the generation of network congestion, also on Resource Allocation in Networks and network resource utilization, well improved and improve;

2) the present invention is owing to adopt the criterion of the utility function of combination multi-parameter as the selection network at portable terminal; Thereby reduced the undue dependence of portable terminal to single environmental factor; More meet the residing overall network ambient conditions of portable terminal; So that the network that portable terminal is selected more to meet oneself requirement is used for professional transmission, can reach the equilibrium of Internet resources and the raising of network resource utilization;

3) the present invention is owing to adopt the network of selecting to be used for transport service with certain probability at portable terminal; All the time select the probability of the network of some optimal utilities through reducing portable terminal; Reduce the network congestion that causes thus; Thereby reduced professional packet loss, improved the probability that professional success is transmitted; Also improved the utilance of a part of suboptimum network, thereby on resource utilization, improved.

Simulation result shows, compares with existing method, and the present invention has improved the probability of success of network resource utilization and professional transmission, has reduced the end-to-end time delay and the packet loss of professional transmission, has realized Network Load Balance, has improved the network service performance.

Description of drawings

Fig. 1 is the basic application scenarios figure that the present invention uses;

Fig. 2 is a routing method for heterogeneous network flow chart of the present invention;

Fig. 3 is the scene topological diagram that the present invention uses;

Fig. 4 is the packet loss analogous diagram that the existing network method for routing obtains;

Fig. 5 is the network success transmission ratio analogous diagram that the existing network method for routing obtains;

Fig. 6 is the end-to-end time delay analogous diagram that the existing network method for routing obtains;

Fig. 7 is the packet loss analogous diagram that the network route method that proposes of the present invention obtains;

Fig. 8 is the network success transmission ratio analogous diagram that the network route method that proposes of the present invention obtains;

Fig. 9 is the end-to-end time delay analogous diagram that the network route method that proposes of the present invention obtains.

Embodiment

For clearly demonstrating the method among the present invention, provided application scenarios figure and flow chart below and be described with reference to the accompanying drawings.

The present invention is applied in the application scenarios of heterogeneous network as shown in Figure 1; This scene is by WLAN WLAN at one; The overlapping region covered of polytype network portion of general grouped wireless network G PRS and wideband Radio Access Network WiMAX; Any one network that current mobile terminal can be selected to have covered itself in professional transmission course is used for transport service, but not is confined to certain particular network; Portable terminal is to find a network that more meets current network ambient condition situation and satisfy self-demand to carry out the business transmission; In whole process, need portable terminal between multiple network, to select one of them network to carry out the business transmission; And constantly predict the situation of network environment state, thereby obtain the optimal path of a transport service.

With reference to Fig. 2, route step of the present invention is following:

Step 1; Mobile terminal period property is obtained current available collection of network Net (i)={ N (j); J=1～n} and corresponding network environment state thereof; Wherein network type comprises WLAN WLAN, general grouped wireless network G PRS and wideband Radio Access Network WiMAX, and the current network that uses of portable terminal is N (i), and the network environment state comprises network availability bandwidth f _Cap, Internet utilization fee uses f _PricAnd message transmission rate rat (N (i), N (j)), handover delay sw (N (i), N (j)) and access delay ac (N (i), N (j)); The initial value of the network sequence number net_num that value of utility is maximum is changed to 0, the time delay estimated value Initially be changed to 30, maximum utility value max initially is changed to 0.

Step 2, portable terminal adopt prediction of Q-learning method and estimation network ambient condition with cognitive ability according to the network environment state that obtains, and according to following formula, obtain the estimated value of each state, obtain required end-to-end time delay T _TotalAnd time delay estimated value

rat_e(N(i)，N(j))＝rat(N(i)，N(j))+λ _rat[rat_e(N(i)，N(j))-rat(N(i)，N(j))]

ac_e(N(i)，N(j))＝ac(N(i)，N(j))+λ _ac[ac_e(N(i)，N(j))-ac(N(i)，N(j))]

sw_e(N(i)，N(j))＝sw(N(i)，N(j))+λ _sw[sw_e(N(i)，N(j))-sw(N(i)，N(j))]

$T_{\mathrm{total}}=\mathrm{sw}\_e(N\left(i\right),N\left(j\right))+\mathrm{ac}\_e(N\left(i\right),N\left(j\right))+\mathrm{pk}/\mathrm{rat}\_e(N\left(i\right),N\left(j\right))+\underset{k\∈\mathrm{Net}\left(j\right)}{\min }\mathrm{Qt}\widehat{\_}{e}_j(N\left(j\right),N\left(k\right))$

$Q\widehat{t\_}{e}_i(N\left(i\right),N\left(j\right))=\mathrm{\λQ}\widehat{t\_}{e}_i(N\left(i\right),N\left(j\right))+(1-\mathrm{\λ})[r_{\mathrm{ij}}+\underset{k\∈\mathrm{Net}\left(j\right)}{\max }Q\widehat{t\_}{e}_j(N\left(j\right),N\left(k\right))]$

Wherein, rat_e (N (i), N (j)) is the message transmission rate estimated value, λ _RatBe the estimating speed of message transmission rate, sw_e (N (i), N (j)) is for to switch to the required handover delay estimated value of network N (j) from network N (i), λ _SwBe the estimating speed of handover delay, ac_e (N (i), N (j)) is the access delay estimated value, λ _AcBe the estimating speed of access delay, the professional size of pk for transmitting,

Be the end-to-end time delay estimated value that business is transmitted through network N (k) from network N (j), λ is the estimating speed of Q value, r _IjFor business is transferred to the value of utility that network N (j) is obtained from network N (j).

Step 3, portable terminal are used and end-to-end time delay network availability bandwidth, Internet utilization fee according to the network environment state that obtains, and carry out normalization through following formula and handle:

$f_{\mathrm{cap}}=\frac{f_{\mathrm{cap}}-{\mathrm{BW}}_{\min }}{{\mathrm{BW}}_{\max }-{\mathrm{BW}}_{\min }}$

$f_{\mathrm{pric}}=\frac{f_{\mathrm{pric}}-{\mathrm{Price}}_{\min }}{{\mathrm{Price}}_{\max }-{\mathrm{Price}}_{\min }}$

$T_{\mathrm{total}}=\frac{T_{\mathrm{total}}-T_{\min }}{T_{\max }-T_{\min }}$

BW wherein _MinBe the obtainable minimum available bandwidth of network, BW _MaxBe the obtainable maximum available bandwidth of network, Price _MinFor network uses the least cost that need pay, Price _MaxFor network uses the costs on the higher scale that need pay, T _MinBe the minimum end-to-end time delay of business transmission, T _MaxMaximum end-to-end time delay for business transmission permission.

Step 4, the normalized value that portable terminal is used according to the end-to-end time delay, network availability bandwidth and the Internet utilization fee that obtain adopts analytic hierarchy process (AHP) to obtain the weighted value that end-to-end time delay, network availability bandwidth and Internet utilization fee are used; The relative influence degree assignment affects intensity grade different, the value of each grade according to the three can the integer within 1～9 in picked at random, the influence degree grade is high more; The numerical value of selecting is more little; Otherwise the influence degree grade is low more, and the numerical value of selection is big more; Suppose that the influence degree grade of picked at random is 1,2,5 first, obtain a comparator matrix W who constitutes by the influence degree grade:

$W=\left[\begin{array}{ccc}1& 2& 5\\ 1/2& 1& 2\\ 1/5& 1/2& 1\end{array}\right].$

Step 5, according to the comparator matrix W that obtains, through | W-λ _MaxE|=0 obtains eigenvalue of maximum λ _Max, wherein E is 3 * 3 unit matrix; Then by coincident indicator formula CI=(λ _Max-n)/(n-1) obtain CI, wherein n is the dimension of comparator matrix W; And according to the coincident indicator CI that obtains, CR=CI/RI obtains CR by Consistency Ratio index formula, and wherein RI is coincident indicator at random, and different n values has fixing RI value, can the coincident indicator table obtains through searching at random.

Whether step 6 can be accepted according to the CR judgment matrix of trying to achieve, and when CR＜0.1, thinks that comparator matrix W can accept, then through Wx=λ _MaxX obtains eigenvalue of maximum λ _MaxCharacteristic of correspondence vector x=(x ₁, x ₂, x ₃) ^T, with the x that tries to achieve ₁, x ₂, x ₃Value is as the weighted value ω of time delay, bandwidth and expense ₁, ω ₂, ω ₃, accomplish the weighted value solution procedure of each state; Otherwise, think that comparator matrix W can not accept, then continue execution in step 3 and reselect the comparator matrix that can accept, calculate weighted value again.

Step 7 according to the weighted value that obtains, adopts the utility function based on multi-parameter, and the effectiveness that comprehensive various parameters are selected route through following functional expression, is selected the maximum network of value of utility and the map network sequence number is stored among the net_num:

Max?U _i＝F _i(f _cap，f _pric，T _total)

F _i(f _cap，f _pric，T _total)＝ω ₁(BW _max-f _cap)+ω ₂f _pric+ω ₃T _total

Wherein, ω ₁Be the weighted value of network residual available bandwidth, BW _MaxBe the total bandwidth of this network allocation, ω ₂Be the weighted value that Internet utilization fee is used, ω ₃Weighted value for end-to-end time delay.

Step 8; According to the maximum network of the value of utility that obtains, portable terminal is in the probability P of [0,1] interval interior real number of picked at random as the selection network; If set up P＜0.9; Then portable terminal is selected the objective network of the maximum network of value of utility as transport service, otherwise Net (i)={ N (j) gets any available network at random as objective network among j=1～n} in set.

Step 9, portable terminal carries out the business transmission according to the objective network that obtains, and judges whether this objective network is identical with current network, if identical, then portable terminal directly uses this Network Transmission professional; Otherwise portable terminal sends handoff request, switches to the objective network from current network, and business is continued transmission through objective network.

Step 10, portable terminal are upgraded network and ambient condition parameter thereof after accomplishing above business transmission again; Promptly the network of the network after switching as the current use of portable terminal; And obtain the new ambient condition parameter under this network, continue to select to be used for the network of professional transmission, when business is sent in the backbone network; Portable terminal is also made the same behavior to the node in the backbone network, to guarantee the optimization of service transmission path.

Effect of the present invention can further specify through following emulation:

Simulated conditions: the simulated environment that adopts MATLAB.

Fig. 3 is the application scenarios topological diagram of the inventive method; This topological diagram representes that portable terminal need send to the destination network from source-end networks with business; Wherein node 1 is a source-end networks, and node 9 is the destination network, and it is professional that portable terminal can freely insert any Network Transmission.Network environment comprises WLAN, WiMAX, GPRS network and corresponding backbone network thereof, and networks of different type has the network availability bandwidth of different data transmission rates, access delay, distribution and Internet utilization fee to use.The current network that uses of portable terminal is wlan network.

Like Fig. 4, Fig. 5 and shown in Figure 6 is in adopting existing network route method; Owing to receive the constraint that resource can not be shared between the networks of different type; A large number of services arrival can make current network be in congested saturation condition for a long time, thereby the business that arrives can't obtain service immediately, causes timer expired; Make packet loss increase sharply; Professional successful transmission ratio descends because of a large amount of packet losses appears significantly in the network, and the business of wait service is also compelled because of network congestion to be stopped etc. to be transmittedly, thereby end-to-end time delay also increases sharply; Simultaneously, because resource can not be shared between the network, uneven professional arrival makes that load is unbalanced between each network between network, and resource utilization is relatively low.

Like Fig. 7; Fig. 8 and shown in Figure 9, the method that adopts the present invention to propose, portable terminal has the cognitive ability to network environment; Can come network is made corresponding effectively estimation and prediction through the sensing network ambient condition; Pass through to adopt the criterion of the utility function of multi-parameter as the selection network according to prediction case then, and select one to be used for the professional network that transmits with certain probability, this network is not necessarily the network of current use.Therefore; When a large number of services arrives; Because the cognitive ability of portable terminal makes the network of selecting can know the congestion situation of network, so business can be by a large amount of obstructions, packet loss has only the trend that increases more by a small margin; The transmission ratio of network success simultaneously also can significantly not reduce because of the arrival of a large number of services, and end-to-end time delay also is just to no longer include than great fluctuation process after increase then is reduced to certain value slowly.It is thus clear that it is balanced that method of the present invention can better realize network resource loads, improves network resource utilization, improves network performance.

Claims (3)

1. the routing method for heterogeneous network based on cognition comprises the steps:

(1) mobile terminal period property obtain current available network type, the sensing network ambient condition, adopt the Q-learning method that the network environment state is estimated to obtain the estimated value of each network environment state, and obtain end-to-end time delay and estimated value thereof:

(1a) portable terminal obtains current available all-network type, with set Net (i) expression, and Net (i)=N (j), j=1～n}, network type comprise WLAN WLAN, general grouped wireless network G PRS and wideband Radio Access Network WiMAX;

(1b), obtain its corresponding network environment state, and obtain message transmission rate, the estimated value of access delay and handover delay through following formula to network type:

rat_e(N(i)，N(j))＝rat(N(i)，N(j))+λ _rat[rat_e(N(i)，N(j))-rat(N(i)，N(j))]

ac_e(N(i)，N(j))＝ac(N(i)，N(j))+λ _ac[ac_e(N(i)，N(j))-ac(N(i)，N(j))]

sw_e(N(i)，N(j))＝sw(N(i)，N(j))+λ _sw[sw_e(N(i)，N(j))-sw(N(i)，N(j))]

Wherein, λ _RatBe the estimating speed of message transmission rate, rat_e (N (i), N (j)) is the estimated value of message transmission rate, λ _AcBe the estimating speed of access delay, ac_e (N (i), N (j)) is the estimated value of access delay, λ _SwBe the estimating speed of handover delay, sw_e (N (i), N (j)) is the estimated value of handover delay;

(1c) according to the above-mentioned estimated value that obtains, obtain portable terminal required end-to-end time delay T when business is transmitted through following formula _Total:

$T_{\mathrm{total}}=\mathrm{sw}\_e(N\left(i\right),N\left(j\right))+\mathrm{ac}\_e(N\left(i\right),N\left(j\right))+\mathrm{pk}/\mathrm{rat}\_e(N\left(i\right),N\left(j\right))+\underset{k\∈\mathrm{Net}\left(j\right)}{\min }\widehat{\mathrm{Qt}\_e_j}(N\left(j\right),N\left(k\right))$

Wherein, Pk is professional size; Pk/rat_e (N (i); N (j)) be business required time of transmission between network,

is that business is transmitted required end-to-end time delay estimated value from network N (j) through network N (k);

(1d) portable terminal adopts the Q-learning method to estimate for the ambient condition that obtains, and obtains the professional end-to-end time delay estimated value that continues transmission from network N (i) through network N (j):

$\widehat{\mathrm{Qt}\_e_i}(N\left(i\right),N\left(j\right))=\mathrm{\λ}\widehat{\mathrm{Qt}\_e_i}(N\left(i\right),N\left(j\right))+(1-\mathrm{\λ})[r_{\mathrm{ij}}+\underset{k\∈\mathrm{Net}\left(j\right)}{\max }\widehat{\mathrm{Qt}\_e_j}(N\left(j\right),N\left(k\right))]$

Wherein, λ is the estimating speed of Q value, r _IjFor business is transferred to the value of utility that network N (j) is obtained from network N (i);

(2) portable terminal is according to the network environment state that obtains, to network availability bandwidth f _Cap, Internet utilization fee uses f _PricWith end-to-end time delay T _TotalCarry out normalization and handle, obtain the weighted value that end-to-end time delay, network availability bandwidth and Internet utilization fee are used, and, select the maximum network of value of utility through following functional expression:

Max?U _i＝F _i(f _cap，f _pric，T _total)

Wherein, F _i(f _Cap, f _Pric, T _Total)=ω ₁(BW _Max-f _Cap)+ω ₂f _Pric+ ω ₃T _TotalBe based on the utility function of multi-parameter, ω ₁Be the weighted value of network residual available bandwidth, ω ₂Be the weighted value that Internet utilization fee is used, ω ₃Be the weighted value of end-to-end time delay, above weighted value adopts analytic hierarchy process (AHP) to try to achieve;

(3) according to the maximum network of the value of utility that obtains, portable terminal is selected this network with probability P=0.9, selects other available network with probability 1-P, as the objective network that is used for professional transmission;

(4) portable terminal carries out the business transmission according to the objective network that obtains, and judges whether this objective network is identical with current network, if identical, then portable terminal directly uses this Network Transmission professional; Otherwise; Portable terminal sends handoff request, switches to the objective network from current network, and business is continued transmission through objective network; Upgrade the estimated value of network environment state and each state simultaneously; When business was sent in the backbone network, portable terminal was also made the same behavior to the node in the backbone network, to guarantee the optimization of service transmission path.

2. the routing method for heterogeneous network based on cognition according to claim 1 is the described network environment state of step (1) wherein, comprises network availability bandwidth f _Cap, Internet utilization fee uses f _PricAnd message transmission rate rat (N (i), N (j)), handover delay sw (N (i), N (j)) and access delay ac (N (i), N (j)).

3. the routing method for heterogeneous network based on cognition according to claim 1, the wherein described state weighted value of step (2) ω ₁, ω ₂, ω ₃, try to achieve through following steps:

(2a) according to time delay, bandwidth and expense influence degree with respect to portable terminal, distribute three's different influence grade, the value of each grade can the integer within 1～9 in picked at random; The influence degree grade is high more, and the numerical value of selection is more little, otherwise; The influence degree grade is low more; The numerical value of selecting is big more, supposes that the influence degree grade of picked at random is 1,2,5 first, obtains a comparator matrix W who is made up of the influence degree grade:

$W=\left[\begin{array}{ccc}1& 2& 5\\ 1/2& 1& 2\\ 1/5& 1/2& 1\end{array}\right];$

(2b) according to the comparator matrix W that obtains, through | W-λ _MaxE|=0 obtains eigenvalue of maximum λ _Max, wherein E is 3 * 3 unit matrix;

(2c) according to the eigenvalue of maximum λ that obtains _Max, by coincident indicator formula CI=(λ _Max-n)/(n-1) obtain CI, carry out consistency desired result with this index, wherein n is the dimension of comparator matrix W;

(2d) according to the coincident indicator CI that obtains, CR=CI/RI obtains CR by Consistency Ratio index formula, and wherein RI is coincident indicator at random, and different n values has fixing RI value, can the coincident indicator table obtains through searching at random;

(2e) when CR＜0.1, think that comparator matrix W can accept, then through Wx=λ _MaxX obtains eigenvalue of maximum λ _MaxCharacteristic of correspondence vector x=(x ₁, x ₂, x ₃) ^T, with the x that tries to achieve ₁, x ₂, x ₃Value is as the weighted value ω of each state ₁, ω ₂, ω ₃, accomplish the weighted value solution procedure of each state; Otherwise, think that comparator matrix W can not accept, then continue execution in step (4a) and reselect the comparator matrix that can accept, calculate weighted value again.

Images