Two-User Interference Channels With Local Views: On Capacity Regions of TDM-Dominating Policies

Abstract

We study the limits of reliable communication in two-user interference channels where each of the two transmitters knows a different subset of the four channel gains characterizing the network state. In order to systematically analyze this problem, we introduce two concepts. First, we define a local view model for network state information at each transmitter, wherein each transmitter knows only a subset of the four channel gains. This subset may be mismatched from that of the other transmitter, limiting the ability of the two transmitters to coordinate transmission decisions. Second, we define a notion of maximal rate regions achievable by transmission policies that dominate time-division multiplexing (TDM). Specifically, these "TDM-dominating capacity regions" characterize rates achievable when transmission schemes must, for every possible realization of network state, achieve rates at least as good as what can be achieved through TDM. We consider a set of seven symmetric local views based on an assumption that each transmitter uses the same mechanism to gather its local view. For five out of the seven local views, we show that TDM is sufficient to achieve the full TDM-dominating capacity region for the linear deterministic interference channel. For these five local views, our result implies that no single policy can achieve a rate point outside the TDM region without inducing sub-TDM performance in another network state. The common traits shared by the two remaining local views (those with better performance than TDM) are: first, each transmitter knows its outgoing interference channel gain, and second, there exists at least one channel gain known to both transmitters. For these two local views, transmitters can use their knowledge to achieve opportunistic rate gains beyond TDM. Using the relationship between the linear deterministic channel and the Gaussian channel, we extend our conclusions to bounded gap characterizations of the TDM-dominating capacity region for the Gaussian interference channel with local views.Index Terms-Compound channel, interference channel (IC), local views, wireless networks.

Robust Signaling for Bursty Interference

“…The converse bound (18) presented in Theorem 4 was derived for local CSIRT, so it applies to the case at hand.…”

“…We study how interference burstiness and the knowledge of the interference states (throughout referred to as channel-state information ( CSI )) affects the capacity of this channel. We point out that this CSI is different from the one sometimes considered in the analysis of ICs (see, e.g., [ 18 ]), where CSI refers to knowledge of the channel coefficients. (In this regard, we assume that all transmitters and receivers have access to the channel coefficients).…”

Two-User Erasure Interference Channels With Local Delayed CSIT

“…A rate-tuple (R 1 , R 2 ) is said to be achievable, if there exists encoding and decoding functions at the transmitters and the receivers respectively, such that the decoding error probabilities λ 1,n , λ 2,n go to zero as n goes to infinity for the given choice of S Tx1 and S Tx2 . The capacity region for View 1 A similar set of local views for the channel state information was studied in [11] in the context of two-user Gaussian interference channel (not fading) to identify the views in which one could outperform TDMA.…”

“…We consider several possible choices for the available delayed CSIT at each transmitter as shown in Fig. 2 (such locality of knowledge was considered in [11] for a different setup). We demonstrate that it is sufficient for each transmitter to only have the knowledge of its outgoing channel gains with delay in order to achieve the same performance of global delayed CSIT.…”

Capacity Analysis and Information Optimization of WSDM

“…However, the guard interval existing in FDM and TDM leads to the loss of multiplexing, which reduces the channel capacity [3] [4]. Since code division multiplexing (CDM) use orthogonal codes, there is a hard limit on how many orthogonal codes can be obtained.…”