bestbarcoder.com

Multiple access in interference in Software Produce Code 128 in Software Multiple access in interference

7.4 Multiple access in interference using software todeploy code 128 code set b on asp.net web,windows application QR Code Spevcification the use of partiall Software code128b y orthogonal channel assignments are means of decoupling the system and allowing distributed algorithms to converge more quickly. Probing is a means of determining how strong the local interference coupling is on a particular channel, and whether alternatives may be better. The question that now has to be considered is how the various adaptive algorithms interact with each other.

Suppose the network contains nodes in which the transmitter and receiver both have adaptive antenna arrays, adaptive equalization, and a wide dynamic range of power adjustment. Adjustment of either the power or the transmit beam pattern alters the interference seen by everyone else, and also modifies the impulse response seen by the intended receiver through selective weighting of multipath components. Adjustment of the receiver beam pattern alters the interference levels and the impulse response, changing the power level required for reliable operation and the residual channel seen by the equalizer.

Thus algorithms are coupled at two levels: internally, with respect to the desired link, and externally, via the interference coupling to other users. There are many possible ways to combine power control, antenna arrays, and adaptive equalizers, not all of which are useful. Closed-form solutions can be derived for the optimal weights of the receiver array given any particular transmitter pattern, but the joint optimization is already difficult for a single link.

The problem is even more difficult when the interferers also have adaptive arrays since, for strongly coupled links (large gains), an action by one produces a large reaction in the others. As there is no closed-form solution to the general form of this problem (and the propagation environment will only be approximately known in any case), a practical approach is to adapt the receiver array using the LMS algorithm (see 6), and then reuse these weights in transmission. The heuristic behind this choice is that, in a time division duplex system, the nulls placed in the direction of the interferers during reception also reduce the interference transmitted towards the other users who caused most of the interference.

Consequently, receiver adaptation also reduces the interference generated. In addition, the gain in the direction of the communications partner is preserved, so that SIR is improved for the link. These two effects also serve to decouple interactions in the network (reducing interference coupling as compared with the primary link), promoting more rapid convergence.

. Example 7.14 Antenn a array adaptation in multipath interference Figure 7.11 illustrates one simple scenario in which users (1,2) and (3,4) are the communication partners, and the barriers A F are highly reflective.

Some of the principal propagation paths are shown, with interference caused only by multipath in this example. All the transceivers possess adaptive antenna arrays for use in reception and transmission. (a) Suppose the link (1,2) is silent, and users 3 and 4 have adaptive arrays but no equalizers.

How should the beam patterns for users 3 and 4 adapt (b) Suppose now that users 1 and 3 transmit in odd intervals and users 2 and 4 in even intervals. How will the antenna arrays behave Solution (a) Users 3 and 4 should adapt their arrays to suppress the multipath off barriers C and E and put high gain in the direct connection, or, if the multipath delay spread is small, the weights should be adjusted so that it adds in phase. A least squares algorithm provides the required compromise that maximizes SNR.

. Multiple source estimation E 1 D 3 4 F Figure 7.11 Adaptiv Code 128A for None e antenna problem with multipath and interference. (b) During reception, user 3 also must suppress the interference from user 2, while user 1 need only be concerned by the multipath from user 2 off barriers A and B.

User 2 must suppress both multipath from its partner and multipath interference from user 3. If user 2 were to succeed in pulling out the interference from user 3 and reused the same antenna pattern during transmission, then user 3 would experience very little interference from user 2 and can concentrate on multipath suppression (or combining)..

An interesting empi rical result is that when users in the same cell are prohibited from using the same channel, the above algorithm always converges. By restricting channel access in this way, it is ensured that the dominant factor in adaptation is the desired link rather than any particular interferer. Thus, such restrictions decouple the system, permitting rapid adaptation by gradient descent methods.

This could be automated to some extent by using a channel assignment algorithm such as the LIA. When the restriction is released, there are many instances in which adaptation fails to converge, which is hardly surprising as there are no prior expectations that the error surface is quadratic. Stability can also be ensured using a mechanism such as voluntary drop-out when insufficient progress is being made in SIR levels.

This mechanism also decouples the system, leading to stable convergence by the remaining users. Indeed, every distributed iterative procedure based on gradient descent fails in some circumstances in multiple access systems, so there must be some non-linear rule that either limits coupling in the first place, or makes a decision to stop adaptation by some subset of the users based on their relative priorities for access. That is, while in most situations gradient descent methods appear to perform adequately, the system must be engineered so that exceptions are rare, and must include a procedure that determines when one of these exceptional circumstances is about to occur so that a back-up procedure can be used (e.

g., try another channel, go to a non-adaptive mode, or both). Put another way, one set of rules is applied to deal with the main part of the probability distribution of problem types, another set is applied to the (pathological) tails, with some effort spent in recognizing what type of problem is present (e.

g., by using probing or a failure recognition mechanism). This is, of course, a fairly standard method of attack in engineering, with the twist that systems may be designed so that the pathologies are made rare.

.
Copyright © bestbarcoder.com . All rights reserved.